Minimum Wage. Inflation and Unemployment;
A Simultaneous Equations Analysis
by
Al1echi M'Bet
A Dissertation Submitted in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy
in Economies
at
The University of Wisconsin-Milwaukee (USA)
December 1984
Minimum Wage. Inflation and Unemployment;
A Simultaneous Equations Analysis
by
Allechi M'Bet
A Dissertation Subm1tted in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy
in Ec onomic 5
at
The University Wisconsin-Milwaukee (USA)
December 1984
f~~~
/i'~ J' /"jJf-
Major Profes50r
Date
W. k~
3'2'i"~S"
Graduate School Approval
Date
Abstract
Minimum Wage. Inflation and Unemployment;
A Simultaneous Equations Analysis
ay
Allechi M'8et
The University of Wisconsin-Milwaukee, 1984
Under the Superv1sion of Professor Richard Perlman
More than fort y years after its establishment, the minimum wage
(henceforth MW) legislation still remains controversial.
Previous
research has addressed this controversy by examining various
employment/unemployment effects of the MW increases over time.
Recent
studies have reached opposing conclusions concerning the employment
effect of the MW.
On the one hand. the MW is reported to have a
substant1al 1mpact on employment. espec1a11y on teenagers.
On the
other hand. it 1s shown that the ~ effect washes out when the
changing compos1tion of the labor force is taken into account.
The
col1ect\\ve evidence from this research provides a broad accounting of
the d1fferent effects of the ~ on d1fferent groups.
unemployment stems in large measure From the tremendous increase in
the female labor supply.
The MW shows an ambiguous impact on youth
employment and wage inflation has no notable effect.
These more accurate esti~tes will lead to policy recommendations
that are more adequate.
}}-<..:I. S/fsr
,
Major Professor
Oate
i i i
Acknowledgements
1 would like to express my gratitude and apprec1ation to all the
members of my doctoral committee for their constant support and
availab111ty in the successful completion of this dissertation.
1 am especially indebted to Professor Richard Perl man for his
very careful guidance throughout my work, and to Professor Elliott and
Dav1d Rogers for their guidance on the empirical investigation.
My special thanks to Dr. Jacques Pegatienan-Hiey who first guided
me towards gradua te studies in the USA.
Financial support from Dr. Achi Atsain, Director of Cires* is
grateful1y acknow'edged.
This doctoral work would not have been
financially bearable without his support.
1 am except10nally thankfu' to the Koppers-Thiem Company for
financial assistance in the form of scholarships.
1 greatly appreciated the continuous computer assistance provided
by Ron Penl.
1 thank Ms. Cathy Nelson for her outstanding editing and word
processing contribution and ability to read my poor handwriting.
1 am grateful to the Rhodes ln Milwaukee and the Mezikofskys in
Boston, for treat1ng me as one of their own during my stay in the USA.
A very special thanks goes to my family for their patience, love
and support.
*Centre Ivoirien de Recherche Economique et Sociale (Center for
Economie and Social Research - Ivory Coast).
iv
TABLE OF CONTENTS
Page
Acknowledgements
iv
List of Tables
v
List of Figures
xi
List of Appendices
xii
Introduction and Statement of the Research
l
CHAPTER
J.
Literature Review
.
3
lime-Series Studies of the Minimum Wage Effects ..
3
Cross-Section Stud1es of the M~nimum Wage
Effect'
,................
4
Theof'"etical Framework. .•••••..•......•......•.•...
8
Covered and Uncovered Settors and the
Minimum Wage .••.....•..••.••......•......•....
14
Special Case of Teenage Employment and the
Minimum Wage
17
II.
Model and Methodology
(a)
A Probablistic Choice-Theoretical Madel of
Employment:
The Theoretical Ambiguity of
the MW; a Two-Sector Model
23
(b)
A New Approach:
A Model of S1multaneous
Equations...................................
31
v
CHAPTER
III.
Sample; Oata and Results
,_ ....•.••..........
51
Part 1:
Employment Equations
Section 1 - Basic Equation:
OLS, GlS.
Log OLS and Log GLS ...•....••••....••••..•.•
54
Section 2 - Extended Basic Equation:
FLS and WI ••.•••••...••••••...•••••..•••••••
61
Section 3 - $imultaneous Equations Estimation
of the Extented Mode 1
65
Part II:
Unemployment Equations
Section 1 - Bas;c Equation:
DLS, GLS. log OLS,
Log GLS
81
Section 2 - Extended Basic Unemployment Equation:
FLS and WI
84
Section 3 - Simultaneous Equations Estimation
of the Extented Model
87
Part Ill:
Labor Force Participation Equat'on
Section 1 - Bas;c Equation OLS. GlS, log OLS,
Log GLS
93
Section 2 - Extended Basic Labor Force Participa-
tion Equation:
FLS and WI ..
96
Section 3 - $imultaneous Equations Estimation
of the Extented Model
99
IV.
$ultlTlary of Results and Policy Implications
106
APPENDICES....................................................
119
BIBLIOGRAPHY
146
vI
LIST Of TABLES
Table
Page
1
Su~ry of selected empirical studies on the impact
of the minimum wage:
time-series and cross-
sectional
.
7
2
Minimum wage legislation in the United States,
193B-19BO .........•.••••......••.••....••••••.•.••.•
12
3
Employment status of white and blac~ youth,
1955 and 197B
.
20
4
Comparison of the direct and total inflation
impact of a 10 percent increase in the level of
the minimum wage in various studies
.
3B
5
List of variables used
52
Estimated impact of an increase in the minimum wage
on teenage employment; basic equation OLS linear ....
55
7
Estimated impact of an increase in the minimum wage
on teenage employment basic equation; GLS linear ....
5b
B
Elasticities:
the estimated effect of a 10 percent
increase 1n the minimum wage on teenage employment
57
9
Estlmated impact of an increase in the minimum
wage on tee nage employment; basic equation:
OLS logarithmic
.
57
10
Estimated impact of an increase in the minimum
wage on teenage employment:
basic equation,
GLS l oga ri thm1 c
.
5B
11
Elasticities:
estimated effect of a 10 percent
increase in the mln1mum wage on teenage employment;
various specificat10ns and funct'onal forms
.
vi i
Table
Page
12
Effect of an increase in the minimum wage on
teenage employment in the presence of female labor
supply OLS 11near
.
61
13
Basic equation extended to the female labor
supply and wage-inflation:
OLS linear
.
62
14
Extended model in the presence of female labor
supply and wage-inflation:
GLS linear
.
63
15
Estimation of the female labor supply:
OLS linear
66
16
Regression coefficients of wage-inflation:
OLS l inear
.
68
17
Estimated impact of an increase in the MW,
FLS and WI on teenage emp l oyment:
2 SLS
.
69
18
Estimated effect of an increase in the MW,
flS and WI on teenage employment:
2 SlS - GlS
.
11
19
Estimated effect of an increase in the MW,
FLS and WI on teenage employment:
Instrumental
variables estimation
.
73
20
Estimated effect of an increase ln FlS, WI and
MW on teenage employment; instrumental variable
method - GlS
.
73
21
Coefficients of elasticity:
estlmated impact
of a 10 percent increase in FLS, WI and MW on
teenage employment:
summary
.
75
Estimated effect of an increase in the minimum
wage on teenage unemployment rate:
Basic OLS
l;near
.
82
23
Estimated effect of an increase in the minimum
wage on teenage unemployment rate:
Basic GlS
1inear
.
82
vi i i
Table
Page
24
Estimated impact of an increase in the minimum
wage on teenage unemployment rate:
Basic OLS
logarithmic
.
83
25
Estimated impact of an increase in the minlmum
wage on teenage unemployment rate:
Basic GLS
logarithmic ........................................•
83
26
Effeet of an increase in the MW on teenage
unemployment in the presence of female labor
supply and wage-inflation:
Extended model
OLS linear
.
84
27
Estimated effett of an increase in the MW on
teenage unemployment rate in the presence of
female labor supply and wage-inflation:
GLS l1near ,
.
85
28
Estimated effect of an increase 1n the MW, FlS
and WI on teenage unemployment rate:
OLS
logarithmic
.
85
29
Estlmated effect of an increase in the MW, FlS
and WI on teenage unemployment rate:
GlS
logarithm;c
.
86
30
Estimated effect of an increase in the MW, FlS
and WI on teenage unemployment rate:
2 5lS
.
87
31
Estimated effect of a joint increase in the MW,
FlS and WI on teenage unemployment:
2 SLS-GLS
88
32
Estimated impact of an increase ln the MW, FlS
and WI on teenage unemployment:
2 SlS logarithm;c
88
33
Estimated impact of an increase in the MW, FlS and
W! on teenage unemployment:
2 SlS logar1thmic -
GLS ••••••.......••.•.....................••..•••.••.
89
34
Coefficients of elasticity:
estimated impact of
a 10 percent increase in FlS, MW and WI on teenage
unemployment:
Summary of results
.
90
ix
rable
Page
35
Estimated impact of an 1ncrease in the MW on
TlFP:
Basic rnodel - OLS l1near
94
36
Estimated impact of an increase in the MW on
teenage labor force participation:
Basic model -
GLS l inear
94
37
Estimated impact of an increase in the MW on TlFP,
Basic model:
OLS logarithmic
95
38
Estimated impact of an increase in the HW on TLFP;
Basic Model - GlS logarithmic
95
39
Extended basic mode':
Estimated impact of an
increase in the MW. FlS and WI on teenage labor
force participation rate - OLS linear
96
40
Extended basic mode':
Estimated impact of an
increase in the MW, FlS and WI on tee nage labor
force participation rate:
GlS linear
97
41
Estimated impact of an increase in the HW, FlS
and WI on teenage labor force participation rate:
OLS logarithmic
97
42
Estimated effect of an increase in FlS, WI and MW
on teenage labor force participation:
GlS
logarithmic
,.............
98
43
Estimated impact of an increase in the MW. FlS and
WI on teenage 1abor force participation:
2 SlS .....
99
44
Estimated impact of an increase in the MW, FlS and
WI on teenage labor force participation:
2 SlS -
GlS method
100
45
Estimated impact of a joint increase in the MW. FlS
and WI on teenage labor force participation:
2 SlS -
10garithmic .,
,
100
46
Coefficients of alasticity:
Estimated impact of a
10 percent increase in the HW, FlS and WI on teenage
labor force participation:
Sunmary of results .....
101
x
LIST OF FIGURES
Figure
Page
1
Graph1cal representation of an effective mlnlmum
wa 9e
.
9
2
Effect of a minimum wage on covered and uncovered
sector
14
3
The disemployment effeet of the minimum wage in
a dynamic setting
18
4
Transmission of the minimum wage effect to wage/
priee inflation
37
xi
LIST OF APPENDICES
Appendix
Page
A
Teenage Employment Equation; Regression Results.
OLS linear; Different Specifications ..••...........
120
B
Teenage Employment Equation; Regression Results.
GLS linear.
Different Specifications
.
124
C
Teenage Employment Equation Regression Results.
OLS logarithmic.
Different Specifications
.
128
D
Teenage Employment Equation.
Regression Results.
GlS logarithmic.
DHferent Specifications
.
131
E
Teenage Unemployment Equation.
Basic + Pcwel
and Basic + EDP
.
134
F
Extended Teenage Unemployment Equation.
Simultaneous EQuat ions
.
136
G
Extended Teenage labor Force Participation
Equation.
Sirnultaneous Equations Estimation.
Instrumental Variables Method.
Different
Functiona l Forms .................•.................
140
H
Results of Covariance Procedure:
Correlation
Coefficients Matrix and Mean, Standard Deviation
of TEMP. FLS, WI, HW, TFLP and THUR
.
144
xii
1
Introduct~on and Statement of
the Research
Numerous studies have emerged over the past decade dealing with
the effects of the min1mum wage legislation on employment (Mincer,
197&; Gramlich, 197&). with special emphasis on employment of
teenagers.'
These studies un1formly show that sorne amount of
disemployment results From the imposition of an effective minimum
2
wage.
Nevertheless. there ;5 considerable dispute concerning bath the
identification of affected groups as well as the magnitude of these
effects.
For example. Marvin Kosters and Finis Welch conclude that
increases in the mlnimum wage would help blacks out of poverty.
M11ton Friedman to the contrary believes that the minimum wage
does blacks the most harm out of any group.
Yale Brozen argues that
minimum wage incre~ses would 1ead to teenage job 1055.
The main objective of this research 1s to provide new emp1rical
ev1dence on the minimum wage effect by extend1ng and amending previous
studies so as to account for the additional1y relevant effects of
wage-1nflation and the continuous increase in the female labor force.
A system of simultaneous equations is employed in contrast to most
prev10us slngle-eQuation studies that utilized either time-series or
cross-sectional data.
Chapter 1 presents a literature review of the minimum wage effect
on employment/unemployment.
Chapter II describes a model of three
2
s1multaneous equations to be used as an alternative specification to
estimate the minimum wage efrect.
Chapter III contains a discussion
of the data and the emp1r1cal r1ndings.
Policy implications of the
study are analyzed in Chapter IV.
3
CHAP1ER ]
LllERA1URE REV]EW OF MINIMUM WAGE EFFEC1S
ON EMPLOYMENT/UNEMPLOYMENl
The MW literature does not of fer consistent or reliable
conclusions.
While several studies of the employment/unemployment
effects of the MW have been conducted. especially the effects of the
HW on teenage labor force status, no settled results are available.
Part of the problem appears to be methodological.
Nearlyall
previous studies used single-equation models and found a negative
employment impact of the MW.
Although the research is consistent in
finding sorne employment reduction associated with the HW increases,
the estimated effects on unemployment are more varied.
Of 1nterest
are studies by Kaitz (1970), Moore (1971) Adie (1971), Lovel (1972),
Mattila (1979), and Charles Brown, Gilroy and Kohen (1981).
However.
a distinction must be made between time-series studies and
cross-section ones. as we11 as between early studies and more recent
ones.
Time-series studies re1y on differences over a period of time to
estimate the MW effect; i.e .. how does youth employment change when
the MW changes.
Most t1me-series research estimates the effect of the
MW on1y for youth.
This group is often disaggregated by age (1&-17:
18-19 and 20-24 years). sex and race.
Peter Mattila (1978 and 1979)
4
and Ragan (1977, 1919) further d;saggregate by school enrollment
status; Gramlich (1976) breaks down the total teenage population by
full-time and part-t'me status wh11e Welch (1976) considers the
distribution of teenage employment by major industry.
(A more
complete inventorv of time-series studies is presented in Table 1).
Cross-section studies:
An alternative approach to the
time-series method is to re'v on cross-sectional data in arder to make
compar1sons between states or metropolitan areas that differ in the
importance of the MW.
Table l contains the major cross-sectional
studies.
The crucial question confronting the cross-sect'onal
approach is to know how to identify differences in the degree of
importance of the MW wh en a single federal HW law applies to all
states.
Statistically. 'ndeed, if the HW remains constant across
states, one cannot estimate the MW effect.
Two methods to overcome this problem have appeared in the
literature.
The f1rst method is the standard direct approach; 1t
examines the l1nk between tee nage employment and a MW variable (Ragan.
1977).
Second, we have the indirect approach orig1nated by Ragan
(op c1t.).
It consists of a two-step technique:
step l tests whether
HW raises youth wage rate in absolute terms and relative to other
wages.
In step 2 the relat10nship between youth wages and youth
employment 1s tested.
This indirect procedure provides another test whether an increase
in the HW reduces youth employment w1thin a cross-sectional setting.
5
4
5
Arnold Katz
and Alan Fisher
use such an approach to analyze the
relat'onsh1p between youth wages and youth employment.
The major
drawback of cross-sectional setting 1s that 1t does not enable us to
study the impact of changes in the Federal MW.
Another distinction is made in the literature on the MW effect
between early studies and more recent studies.
On the one hand, mast
early studies use 1960 census data and ask whether state minimum wage
laws lowered teenage employment.
Unemployment eQuations were a characteristic of those early
stud1e5.
They assumed that teenage employment/unemployment was only
affected by the current value of the MW. althouqh some allowed for
lagged response.
Kaitz (1970) used data From the period 1954-1968,
while Moore (1911) covered the period 1954-1958.
lhe more recent studies. on the other hand, have estimated the MW
effect on the employment/population ratio and labor force
partlcipation rate. and have derived the unemployment effect from
these (Mincer, 1976; Welch. 1916; Brown et al., 1981).
8ut due to the
extension of the Federal MW coverage in the 1960's, the importance of
state laws has been reduced; and later research analyzed mainly the
impact of the Federal MW on the labor force status.
lhese recent studies find a 10ss in employment which ;s often
offset by the dec11ne in labor force part1c1pat10n so that only part
of the reduction in employment ;s attributable to an increase in
6
unemp10yment .
6
Table l presents a summary of the find1ngs of selected emp1r1cal
studies on the impact of the MW; for bath time series and
cross-sect1onal data; and early studies and recent research as wel1.
)
rob'• ,
SUlftMIry of Selected Upirical Studles on the Inpoct of the *:
Ti_ -series ond Cross-sectlonol
TI HE-$ER 1ES STUO 1ES
[ff&et of an lncnBasa
.... ,0'1
ln the Minimum Wage
uthorlOate
Co......
Dependent
endlor Coyerage on
CoYarage
f Publication
bV Study
Variable
Dependent Ver 1abl e
Includlld
Kaitz (1970>
'954-<>8
Aggr9g6te Teenoge Eq:llo'fll8l1t
Inconcluslv.
V••
and Unempl~nt
J'bore (1911)
1958-69
Aggregote Teenage U~
Incr90ses 1eenoge
V..
ploymeot
Unetrp 1ovment
die 1\\971)
1950\\-70
Aggregate Teenege Unern--
1ncreases Toenage
Ves
plo.,.ment
UfIeI!1) 1oyment
ove 1 (1972>
1~4-70
Aggreg6t& 1eenage Unem--
Inconclusive/Ho
No
plovment
Change
Welch (1974)
'954-<>8
Teenage E1Pt:IJovment in
Decrease Teenage
V..
""nufacturi ng, Ret. fi Trede
Emplovment
and Sel-vlce Sector
rami ich (1916)
'954-<>8
Aggregote ElTlJlov-n+1
Decreas8 labo.. Force
Ve'
population and Aggregate
and ~Ioyment fo,
labo,. Force/Papu t oti 00
!h)st groups
for 10 age/sax/race groups
Ragan (1917)
196}-72
Aggregete ~ lovmeni-l
Decrease Labo,. Force
V..
Population and labor force!
end ~ 1o.,.ment fo,
Population bV age. se)(, race
Teenagers
Marti 1" (1978)
1947-76
Aggregate Employment/Popula-
Decrease Eq:llovment .Ml
V.s
t 1on and Aggregate labor
labor force
force/Population for age and
se:oc groops
.,.
rown et
1954-79
UI'IeII'p 1oyment/PopIJ lat 1on
Decrease labor Force
V..
(1981)
and labor force/Population
and i ncréase Teenage
for Teenagers
Unempl~t
CROSS-SECTlaNAl STUOIES
~lch and
1 1970
Aooregatu Employmentl
Decrease Teenage
Vos
unningh511 (I978) State
Popu\\aton by age groops
~l~t
hrenbergl
197D
Youth Eq:lloyment~opu1atl on
1ncreoS41 Wh 1tu Youtt\\
V..
Ma':cus
Statu
ratio by enrolllNltlt Status
~Ioyment.
Not r&-
( 1979)
ported for non-wh i tu
yeuths
" .....n
197D
Eq:llovment-Populatlon ratio
DecreaS41 labor Force
V..
( 1979)
SMSA
and labor force Participation
Participation and em-
and ul'IeII'ployment rate
ployment 1ncrese unen-
ployment
.
Source:
Adapted frem Charles Brown et al. (1981), p. ZIB.
Mlnunurll Wage Study eonmlssion. Vol. 6.
B
Theoret1cal Framework:
Academie discussions of the MW start with the implications of
standard ecanomie theory for the effeet of the MW on employment.
The
objective is to explain and to test the predictions of the theory.
Historically, the purpose of the Fair Labor Standards Act (FlSA) of
1938 h
to raise wages "without substantially curtailing
employment.~7
According to theory. the underpinnings of HW research lie in the
traditional Marshal1ian setting.
This conventional supply and demand
mode' depicts the eQuilibr;um waqe and employment levels at Wo and [0
respectively as in Figure 1.
With 'the imposition of a mandatory M.rl
(Wrn). the theory predicts a decrease ln employment From OEo to DE,
and a labor surplus of [,-E2" as a result of a fall in the labor
demand and a rise in the quantity of labor suppl y at the new wage (Wrn).
Wage
SL=Supply
of
labor
Potent1al unemployment gap
Wo
_
Dl=Demand for
labor
I--"""------
...,jl.,,
-'p Emp 1oyment
o
Ligure 1:
Graphica1 representatlon of an effective MW effect.
10
At the above-equilibrlum wage (Wm). employers hire few workers or
hire the same number of workers fewer hours per week.
However. the
theory does not specify the magnitude of the reduction in employment
or the work week..
Still at WIn, fewer jobs (El) are available for more
workers (E2) and not all wil11ng. ready and able to work at the MW
will find jobs.
The model holds the gap El-E2 as a measure of maximum
unemployment and the ratio (E1-[2)/(0-[2) as the unemployment rate.
Obviously not all of the E1-E2 workers will be counted officially as
unemployed; but the amount of past research focusing on the
unemployment effects of the MW is a testimony to the popular1ty of
this 1nterpretat1on.
Two major criticisms of the basic theory of the competiti~e labor
market appear in the literature.
First, in labor markets characteriled by monopsony, a skillful1y
set MW may actually increase employment.
lhe monopsony model has not
moti~ated recent work. perhaps because there is little e~idence that
it is important in modern-day low-wage labor markets.
A second l1ne of crHicism concerns the "shock" argument.
lhe ·shock n
argument is the one in which the employer is able to obtain greater
levels of effort in response to the MW increase.
Brown et al. note
that if employers do not minimize costs. there is a possibility that
they wlll respond to the HW increase by raisinq the productivity of
their operation. to offset the MW increase.
lhis offsetting operation
labelled "shock" effect. might reduce the disemployment from the MW,
not e11minate it.
11
It is difficult to test the valid1ty of the monopsony model and
to measure the ·shock· effect.
As a consequence, the empirical work
has re11ed on the basic supply-demand model.
The theory. however. has
been refined in four ways.
First as an extension of the basic model. bath the covered and
uncovered sectors were introduced (Welch. 1974; Gramlich, 1976~
Mincer, 197&).
In fact, the FlSA even with its mast recent amendments
in 1977, is not universal but includes certain industries engaged in
intrastate commerce and a11 industries engaged in interstate
commerce.
About 84% of al1 private nonfarm. nonsupervisory wage and
low~age workers have been subject to the MW 1n 1978 compared with 53%
in 195D (Welch. op cit., p. 3).
lt is therefore important to consider
a model in which coverage is complete and another where it is
incomplete.
Table 2 puts forth the evolution of and the coverage by
the mlnimum wage.
Theory suggests that the imposit;on of the MW reduces employment
in the covered sector.
Workers unable to f1nd jobs in the covered
settor will either (1) work in the untovered settor; (2) withdraw from
the labor force; or (3) remain unemployed w1th the expectation of
gett1ng a job in the covered settor.
Consequently, a flow of workers
will occur between the two sectors and this will eventually shift the
supply curve in the uncovered sector if alternative (1) ;s chosen
(i.e., the unemployed workers choose ta work in the uncovered
sector).
A downward pressure will be put on the wage in that settor.
Table 2:
Minimum Wage leghlation in the UnHed States. 1938-1980.
Effecthe
Minimum Wage Relative to
Date
Percent of
Aver.MLHou~age ln Manufacturing
of M1n1mum
Nom; na 1
Nonsuperv1sory
Wage Change
Minimum Wage
Employees Covered
8efore
After
10/24/38
SO.25
43.4
0.403
10/24/39
0.30
47.1
0.398
0.478
10/24/45
0.40
55.4
0.295
0.394
1/25/50
0.75
53.4
0.278
0.521
3/1/56
1.00
53.1
0.385
0.512
9/3/61
1.15
62.1
0.431
0.495
9/3/63
1. 25
62.1
0.467
0.508
9/3/64
l. 25
62.6
211 /67
1.40
75.3
0.441
0.494
211 /68
1.60
72 .6
0.465
0.531
2/1/69
1.60
78.2
2/1/70
1.60
78.5
2/1/71
1.60
78.4
5/1/74
2.00
83.7
0.363
0.454
111/75
2.10
83.3
0.423
0.445
1/1/76
2.30
0.410
0.449
1/1/78
2.65
0.430
0.480
1/1/79
2.90
0.402
0.440
111 /80
3.10
0.417
0.445
1/1/81
3.35
Source:
Ehrenberg/Smîth (p. 69).
-N
13
But as Hughes and Perlman (1984) observed, if wages in the uncovered
sector are not assumed to be fal11ng freely because of what Reder
(1955) called a ·social minimum," then it can be the case where wage
rate rises in bath settors.
Figure 2 presents a graphital analysis of the unemployment
effects of minimum wage in a two-settor model.
Before a MW is imposed, wages are equal in bath settors.
An
effective MW raises the wage level in the covered settor to w~.
We assume new entrants into the labor force and flexible adjustments
of workers in bath settors.
As a result of the h1gher MW, an
unemployment gap of AC 1s created in (c); AB represents the displaced
workers in (c) and BC the transfer of workers From the uncovered
market. attracted by the new higher minimum wage.
Thus to the work force E~ formerly in the uncovered sector,
are added the displaced workers E~ - E~.
But because the
demand curve for labor slopes down in the uncovered sector also. the
increased supply of potential wor~ers drives down the wage From
WU to Wu.
o
In the covered sector, workers are selected positively on the
basis of their marginal productivity; in the uncovered sector however.
getting a job will be negatively related to s~ill.
Such partial
coverage of the minimum wage produces winners and losers.
The winners
are those wor~ers in the covered sector who keep the1r job ex post;
the losers are those low-skilled workers who lose their jobs in the
covered sector and now are pa1d a lower wage in the uncovered sector.
Hence on balance, there 1s no impact on unemployment.
But there are
redistributional effects in the process.
14
w
Covered Sector (c)
Uncovered Sector (u)
SI.
s..
'Wu
- - - -
o I..-.....L_ _..L,-
'!>"
E,c.
~
E. v
f.V
E
1
1
F19ure 2:
Effects of a Minimum Wage on Covered and Uncovered Sectors.
15
lhe second ref1nement of the basic model 15 to redef1ne who are
the unemployed El-E2 in Figure l by separat1ng ·d1scouraged workers·
from the unemployed.
D1scouraged workers want a job but have given up
searching and are classif1ed ·out of the laber force- and excluded
from the official unemployment count.
This expla1ns the reason why
the basic model cannat pred1ct unambiguously the effects of the MW on
unemployment.
The third refinernent is in connection with the recent shift to
the use of emplovment as dependent variable instead of focusing on
unemployment as did many prev10us empirical studies.
Brown et al. (1981) argue that the employment 105S is a better
measure of the "harm" done by a rise in the MW than is the change in
unemployment.
They note that because of the Rdiscouraged" workers,
the harm measured by the change in unemp10yment is understated.
"incer (1976) finds the discouragement effect to be stronger for
many demograph1c groups with a reduction in the 1abor force
participation of the affected workers.
M1ncer ' s find;ngs suggest that
the unemployment increase is less than the employment decline.
"'elch (1974) supports the study of the more dennite empl.oyment
effect:
~because of the ambiguity of the standard mode1 concerning
the effect of "Won unemp1oyment, it is surpr;sing that the majority
of empirical analyses of MW effects have focused on unemp10yment
rather than on emp10yment, where predictions are unambiguous. at 1east
for competitive 1abor markets. N
1&
Several reasons eKpla1n the persistent use of unemployment as
dependent variable.
F1rst there is a natural tendency for policy
purposes to want to fDeus directly on what ;5 seen as the problem:
unemployment with al1 lts soc1al and political ramifications.
The
second reason 15 the reluctance to part with the attractive simplicity
of the basic supply-and-demand mode'.
But analysts have s1nee come to
realize that 1ncent1ves to withdraw from the 'abor force have been
increased by the ava11ability of second level opportunities such as
school enrol1ment. welfare, non-market work. and armed forces.
Furthermore. focusing on employment status allows the distinction
between ful'-time and part-time employment.
In addition. the changes
in the method of measuring the labor force status introduced to the
Current Population Survey (CPS) in 19&7 affected the CDunt of
unemployed s1gnificantly more than the employed (Steln, 1967; Summers,
1981).
Summers describes some of the movements in the unemp10yment
rate as spurious Rbecause of the uncertainties surrounding the
statistica1 procedures used in measuring unemp1oyment.-
Response
errors, sampllng errors and seasDnal adjustment errors create biases
in the unemployment rate.
Moreover, inconsistencies of people
intervlewed across rotation groups, undercoverage. nonresponse and
even noninterviews create great standard errors assoc1ated with
unemp10yment and employment rates.
The standard errors in the
unemployment rates are greater ref1ecting the sma11er sample size; the
errors in employment-population ratio are smaller because this rat10
;s a much larger number.
Summers then suggests the changes in
17
employment as alternative labor market 1nd1ctators.
He concludes:
·Changes in employment, measured by the establishment survey. May
prov1de a better guide to changes in labor market conditions than
changes 1n unemployment because of ambiguities in the definit10n of
the labor force. '1
A final refinement of the basic model can be cons1dered under a
dynam;c settinq as it appears in Figure 3.
When the overall trend in
output is upward, the theory suggests that rather than necessarily
reducing employment. the MW would retard the rate of employment
growth.
As in Figure 3, the initial equilibrium is at Wo and Eo for
wage and employment respectively.
But as the demand for labor is
increasing as ecanomie activity expands, a minimum wage Wm is
imposed.
At wm the new equilibrium level of employment is settled at
E instead of at El where Eo < [2 < El.
ln such a case, the MW
2
need not reduce employment ln an absolute sense, but it would reduce
the rate of growth in the employment from what it would have been in
the absence of the MW.
However, the 10ss of potential employment is
counted as employment loss due to the HW.
A fonmal demonstration of the ambiguous effect of the HW on the
labor force status is provided in the next pages.
Heterogene1tY in the labor market:
The special case of teenaqe labor.
Homogeneity in the labor force was the underlying assumption of
the theoretical analysis of the effect of the HW presented ear11er.
In fact, there ex1sts a great deal of disparity among workers due to
lB
SL = Suppl Y of
labor
w()
DL = Demand for
labor
o I - _........
...L-.....L.
~
E '" t'oy.
mc.nt
F1gure 3:
The Disemployment Effect of the MW in aDynamie Setting.
19
characterist1cs such as age. sex, race. human capital. etc.
Indeed
teenagers have in recent years experienced great difficult1es getting
jobs, a fact sorne economists attr\\bute to the Federal minimum wage
legislation.
The overall assessment 50 far, based on theory and past studies.
reveals that the MW reduces employment and. in 50 doing,
generate
sorne redistributional effects.
Concern has especially focused on teenagers who lack work
experience and skil1s.
Sorne adults are also denied jobs as a result
of the minimum wage; but overall teenagers bear a disproportionate
share of the burden.
The table below provides sorne supportive infonmat1on which
compares data for two years, 1955 (well before important changes in
the MW took place) and 1978.
For 1955 Ostenman found that the unemployment rate of 35-44 year
old White males was 2.6 percent.
In 1978, it stood at 2.5 percent.
The two periods are Quite comparable.
likewise White teenagers
have held up Quite well with a small rise in the employment to
population ratio.
The major development during the same period of time 1s the
virtual collapse of the labor market for Black teenagers.
Congress
and successive administrations have shown sorne w1l1ingness to retain
the minimum wage protection for most adult workers.
However. because
of the overr1d1ng teenage unemployment rate (more than double the rate
of adults), sorne analysts have suggested a -dual" or "subminimum· or a
Ntwo-tiered system,· w1th a 10wer minimum wage for teenagers than for
adults.
20
Table 3:
Emp10yment status of White and Black youth, 1955 and 197B.
Unemployment rate
Population Subgroups
1955
197B
16-19 year old Wh1tes
10.3%
13.9%
16-19 year old Blacks
15. B%
36.3%
Employment to population
Populat1on Subgroups
ratio
16-19 year old White male
.52
.56
16-19 year old White female
.37
.49
16-19 year old Black ma le
.52
.30
1&-19 year old Black female
.26
.23
Source:
Paul Osterman (19BO), pp. 115-122.
21
But such special treatment for youth has been opposed by labor
unions, on the ground that employers rnight lay off adults w1th
families to support, in arder to h1re teens at a lower d1fferential
wage.
Furthermore. why not grant other subgroups with problems special
treatment to encourage their employment?
Is such treatment justified
in any dimension--human and otherwise (ecanomie)?
Po11ttcs is known
as the art of the possible; hence it is only natural to concentrate
onels effort on a segment of the labor force where change is
considered probable.
Thus, for the first time. the high teenage
unemployment led to an amendment of the Federal Minimum wage Act of
1977.
It was proposed to set a sub-minimum rate for youths in
Congress.
The amendment ~as defeated by a single vote; this
demonstrates the offsetting strength of the argument for and against a
lower subminimum for teenagers.
like the Nixon Administration ~hich
favored a subminimum ~age for teenage ~orkers, the Reagan
Administration recently proposed l'the 1984 Youth Opportunity Wage Act R
~h1ch ~as introduced in the U.S. House and Senate on "aY·17. 1984.
Congress has debated the issue time and again since 1980 but taken no
action on the bill.
The bill ~ould amend the Fair labor Standards Act
to permit employers to pay youth an hourly ~age of $2.50--i.e. 85
cents belo~ the legal minimum.
Econometrie studies that sho~ a good deal of hanm caused to
younger ~orkers by the MW bolster the case for a subm1nimum ~age.
However. as ~ith most controversial issues, economists can not agree
22
about the true magnitude of the changes brouqht by the MW.
An
examination of selected investigation in Table l demonstrates how
divergent are the estimates reached by d1fferent analysts.
Because of such controversy over the magnitude of the estimates,
and like most prev;ous studies, this dissertation deals with teenage
labor force status as the dependent variable; and we intend to test
the robustness of the degree of the negative impact of the HW on
teenage labor force status, using a new approach.
Preceding such an approach is the presentation of a theoret\\cal
model of employment.
23
CHAPTER Il
MODEl AND METHODOlOGY
A Probabl1stic Choice-Theoretical Model of Employment
The purpose of anticipating the MW effects on a priori grounds is
to penmit a more intelligent evaluation and interpretation of
empirical findings.
But to understand this evaluation fully, certain
theoretical aspects should be made to provide a framework within which
to adequately interpret empirical results.
A full theoretical
treatment i5 beyond the scope of this study. however.
Instead, a
brief out Quite substantial theoretical model ;5 presented that should
explain the ambiguity in empirical results.
The Theoretical Ambiguity of the MW:
A Iwo-Settors "odel
A comman theme of ecanomie literature is that the MW decreases
employment. w1th only the empirical magnitude in doubt.
Actually. the theoretical link between the MW and Employment ;s
not at all straightforward as it appears.
On the contrary. only one
sector model. B i.e .. model with complete coverage, f1nds such
unamb1guous negative employment effect of the MW.
The mode1s of Welch (1974) and Mincer (197&), although more
sophisticated are still incomp1ete.
They do recognize the existence
of the uncovered sector but fail to consider the genera1-equilibrium
repercussions of a change in the MW.
Z4
Hence. instead of using partial-equ11ibrium analysis, or 19noring
the uncovered sector. we present a more elaborate model in a
general-equilibrium framework along with the uncovered sector. 9
Assumptions of the Model
a) Consider the labor market impact of the MW within a two-sector
competitive model, i.e., a partitioned market ;nto the uncovered (u)
and the covered (c) sectors.
h) Labor is homogeneous.
c) Wages in the covered sector are equal to the HW or the market
wage.
d) Wages 1n the uncovered sector are determined by supply and
demand.
e) The HW is effective; that i5, it is set above the market wage.
Under the above assumptions, we show that in a
general-equ11ibrium setting, there exists sorne ambiguity in the impact
of the HW on employment.
Tndeed, conditions exist under wh1ch an
increase in the HW will increase total employment.
To demonstrate
that point, consider an individual deciding to enter the labor
market.
The individual's decision to enter the labor force depends on
the "expectedA utility of market act1vity relat1ve to the expected
util1ty of non-market activity:
leisure and work outside of the labor
market.
The individual maximizes expected ut111ty (instead of
utility) in order to express the fact that the indiv1dual ' s choices
are made under uncerta1nty.
The expected ut11ity of gett1ng a job in
a given sector ;5 determ;ned by the PROBA81LITY of obtaining a job in
that sector and by the expected wage. condittonal on job
25
aval1ability.
Assumption d) states that the uncovered sector clears:
thus in a perfectly competitive world, the probability of finding a
job there 15 unity.
ln the covered sector, however, the probability
of f1nd1ng a job is inversely related to the excess labor supply.
The expected utility of market activ1ty can be expressed as:
(1) E(U)=ma.
E(UC)=f(HW. PB)
E(UU)=f(Wu• 1)
fi > 0 for j=PB
C
where U = utility From entering the covered sector;
UU = utility From entering the uncovered sector;
MW = minimum wage measure
PB = the probability of finding a job in the covered sector
W = wage rate in the uncovered sector.
U
The Supply Function of Labor
Based on (1), ceteris paribus. an increase in the MW, PB, or Wu
will attract more people into the labor force.
The degree of
attractiveness will be higher the higher, the value for HW and PB of
enter1ng the covered sector.
The supply of labor in the covered and
uncovered sectors both depend on function of Wu' HW, and PB; and
theY can be expressed respectively as:
C
L = LC(W
(2)
u, ~. PB)
LU = LU(W
HW, PB)
(3)
u'
The Demand Function for labor
The labor demand in a sector is negatively related to the wage
prevailing in that sector, featuring the downward-sloping demand curve
(2) and (3) are the labor supply functions; they may be affected
by other factors.
Since we focus on the impact of the MW, ceteris
par1bus, we ignore at this theoretical level the values of all other
exogeneous variables for the sake of clarity.
26
for labor; 1t 15 a150 a funct10n of product priee (p) wh1ch in turn
depends on the wage rates in the two settors.
The MW increases the
eosts of labor 10 the covered settor, thereby rais1ng the priee of the
product produced there relative to the priee of product in the
uncovered settor.
Consumers, assumed to be rational and
wel'-infonmed. substitute goods produced in the uncovered settor for
those produced in the covered settor.
Labor demand in the uncovered
settor expands and labor demand in the covered settor contracts.
This
is precisely the SPillover'O effect ignored in partial-equilibrium
analysis as in Mincer (1976).
The direction of such a sp1110ver
effect is not known a priori. leading to the ambigu1ty of the MW
effeets.
After full adjustment, the labor demand ean be expressed as:
De
Oe(MW, P)
(4 )
DU • OU(W , p)
( 5)
u
P is the priee of goods produced in the eovered seetor relative to the
pr1ce in the uneovered seetor; p = p(MW, Wu)
(6)
The model is closed with the following relations:
- PB the probability of getting a job in the covered sector is
inversely related to the e~eess labor suppl Y (Le-OC); hence
PB • PB(le, OC> or alternativelyll
PB =
a.
le
DC+a.
(7)
The term a. guarantees that PB lies between 0 and 1; 0.>0 and O<PB<l.
Neeessary to close the model also is the equality of the supply of and
the demand for labor in the uncovered sector:
Du = lU
( 8)
Also we have the followi ng identities:
N - Ne + NU
(9)
l - le + lU
( l 0)
27
The 1dent1ties (9) and (10) state that total employment (N) Dr Labor
Force Cl) is the sum of employment (labor force) in the two sectors.
Comparative-Statlc Analysis of the Model:
The Effects on
Employment of a Change 1n the MW
From equatlons (2) to (10) the signs of the partial derivatives
are assumed to be as follows:
al C = le
<
0;
= Le
>
0;
al U
lU
<
o
WU
m
aWu
aMW
m
C
al
=
lC
<
o·,
alU = lU
<
0
p
p
aPB
aPB
c
u
ao
OC
< o·,
ao
= OU
>
0
m
p
aMW
aP
~ = Pm > o·, aP
= Pw
<
O.
aMW
aWu
II we combine eQuations:
OC = OC(MW, Pl
(4)
OU = OU(Wu, P)
(5 )
and
N = NC + NU
(9 )
total employment (H) can be expressed as:
N = OC(MW, Pl + OU(Wu, P)
1
Interpretation of eguation 12
The term O~ indicates the direct impact of an increase in
the HW on total employment N.
Recall that by assumption o~<o,
implying that the higher the HW, the less the labor demand in the
covered sector; and as a result employment declines.
The second term OU
aWu captures the effect of the increase
w
aMW
in the MW in the uncovered sector.
The higher wage in the covered
sector will attract sorne workers from the uncovered sector into the
covered one.
At the same time the increased labor supply in the
covered sector lowers the probability (PB) of getting a job there and
hence drives sorne workers into the uncovered sector.
The crucial
point is that once. the spillover effects are considered, the wage rate
(Wu) in the uncovered sector may rise or fall.
Therefore, the sign
of the second term is undeterminate.
A rise in Wu will have a
negative effect on employment and vice-versa.
The third term (Oc-Ou)ap
is the last impact of a change
p
p aHW
in the MW on the employment.
A change in the MW alters the relative
29
product priee (p) and consequently alters labor demend in each of the
two settors.
In partial-equ111br;um analys1s, this overall effett is ;gnored
because we are holding ather things constant that are not in fact
constant.
In 5uth a partial analysis. OC ~ OU ~ O' henee we
p p '
have an unambiguous negat1ve effect of the MW increase on employment
on one hand.
ln general equilibrium on the ather hand. (D~ +
Du) >
0 and therefore the impact of MW on N is undeterminate
p
<
since (OC + Qu)ap
May be either positive or neqative.
If positive.
p
p aHW
this term May exceed the sUm of the first two; then in response to a
change in relative praduct priee resulting from a change in the MW.
'abor demand in the uncovered sector will increase sufficiently to
12
offset the disemployment effect in the covered sector.
Ta summar1ze. in a two-sector competitive model. the impact of a
change in the MW on Employment is theoretical1y ambiguous when the
spillover effects are taken into consideration.
Most previous studies
failed to acknowledge this fact and found inaccurate results.
For
example. by omitting the female labor supply. those studies failed to
examine the most fundamental change in the labor market in the united
States since 1945:
the increased labor force participation of women.
Accord1ngly. previous studies did not answer the crucial policy
question:
Have women been taking jobs away from teenagers?
After
this demonstratlon of the ambiguity of the impact of the MW on
Employment. we can proceed to the empirlcal estimation.
30
The Equations of the Emp1r1cal Model and Estimation Methodology
Uslng Equat10n (11), speclfled as N = OC(MW, P) + OU(Wu, P),
where N 15 defined as Total Employment in bath the covered and
uncovered sectors. one can generalize N by including the unemployment
generated in bath sectors by the imposition of the MW.
N can
therefore be redefined as a measure of labor Force Status (label1ed
Y). i.e., employment and unemployment.
The effects of the MW on the labor Force status can hence be
tested.
Such effects have been studied extensively for youths.
Most previous research used a single-equation model of the form;
y = f(MW, D, Z) where -
-y is the dependent variable and a measure of labor force status,
-MW is a measure of the minimum wage.
-0 represents a business cycle variable; and
-Z is a lSst of the other socio-economlc variables.
Table l (p. 9) presents a surrmary of se1ected empîrical studies.
On balance, it is found that the effect of a 10% increase in the MW is
estimated to result in a 1-3% reduction in total teenage employment
(Brown, Gllroy. ~ohen, 19B1).
Such low results may be due to the
omission of relevant explanatory variables in the equation.
The question arises:
whîch Icontrol l variables, i.e., variables
other than MW, should be included in the estimating equat1ons?
Although there is general agreement that other explanatory
variables should be introduced in the equations, there exists a
comparable wide disagreement as to the appropriateness of
31
1ncorporat1ng supply-side variables in that direction (Ad1e-Gallaway.
1973; Lovell. 1973).
This important issue has yet to be resolved and
the current research is a tentative step in that direction.
Mention was made earlier about the new development that occurred
in the American labor force:
the emergence of a strong female labor
force participation rate.
To the extent that women. especially
married women worklng part-time are likely to be substitutes for
teenagers (Hamermesh, 1919). it is important to add a measure of
women's labor supply in the estimating eQuation.
Another major factor thought to influence the labor force status
is the wage-inflation/price-inflation spiral.
According to Ehrenberg
(1981), "Perhaps the most pressing econom;c prob1em today ;s the
wage-price unemployment issue. Il
Hence a wage-;nflation eQuation must
be specified in the model.
A new approach to the problem:
A model of simultaneous equations of
minimum wage, unemployment, inflation and female 1abor supp1y.
In th1s section, we develop a mode1 that ful1y takes into account
wage-inflation, female labor supply and unemployment.
Equation 1:
The employment/unemployment equation 1s a variant of
the Charles Brown et al. model.
The general fonm 1s Y= f(MW, D, Z),
AH,
where Y 1s as before a measure of labor force status; Y ;s def1ned in
the present study as the teenage employment, unemployment and labor
force partlcipat10n rate alternat1ve1y.
32
- MW 15 a measure of the minimum wage.
- AHE 15 the Average Hourly Earnings measure.
It standard1zes for the
erosion due to rising priees and growing productiv1ty.
- 0 1s the aggregate demand. standing for business cycle variables
which account for changes in the level of economic activity.
- Z is a host of other explanatory variables to control for labor
supply such as school enrollment. participation in the armed forced,
age, sex. race.
For the purpose of comparability, we propose a
modified version of the Brown et al. eQuation. for the same period
1954-1979 with Quarterly data:
y.
60·6,KIMW+B2WI+63FL5+04POP+OSEDP+O&AFP+B7EF7PP+OBPCWEL
+B97+610750+01102+01203+0'304+<
Definition of Variables.
1 - The dependent variable (V) could be defined in several ways.
In
practice, the ratio of employment to population (E/P) is most often
used (Ragan, 1979).
For our purpose, however, we will use alternative definitions
of Y.
Specifically, the teenage unemployment rate (THUR) will be
introduced in addition to the employment population ratio (EMP) for
the youth subgroup employed.
2 - KIMW is the popular Kaitz index of the minimum wage, using teenage
unemployment as
we1ght.
lt is the MW deflated by average hourly
earnings to take into account the fact that the MW is not important
33
per se but 15 important in relation to other wage rates.
In addition,
one must cons1der the (average factor.
The MW should have a greater
influence the greater the number of workers covered by the law.
For
this reason. the MW variable is we1ghted by the fraction of workers
subject to MW legislation.
K[HW
[Hwi
Cil + [HW' . Ci'l
AHEi
AHE
-[ ~ non-agricultural employment
-MW = nominal value of the minimum wage.
-AHE = sorne broad measure of average earnings of nonsupervisory
workers.
For example, average hourly earnings in either manufacturing or
the private economy.
13
-C = proportion of nonsupervisory workers covered by the basic MW.
MW* = minimum wage for newly covered workers.
C* = proportion of workers for newly covered workers.
i = major industry division.
t = total private nonagricultural economy.
3 - WI = Wage Inflation
4 - FLS = women's labor supply.
(Wl and FLS are specified in the next pages.)
5 - POP
Ratio of teenage population to total civ111an population.
6
ratio of teenagers enrolled in school to the teenage
civi11an population.
1 - AFP = The ratio of teenagers in the armed forces to the total
teenage population.
34
8 - EFTPP = The ratio of enrollment in Federal training and employment
programs of those aged 16-21 to the civillan population aged 16-29.
9 - PCWEL = Pr1ce-deflated Nwelfare" aid to fam111es w1th dependent
children. food stamps and commodity distribution programs benefit per
wornen of child-bear1ng age (16-44).
la - T = A l1near time trend.
This time trend is included to control
for the impact of technological change on firms' demand for teenage
labor.
11 - TSO
lime sQuared.
12. 13, 14 ~ 02. Q3. Q4 respectively are dummy variables for the
second. th1rd and fourth Quarter.
The first Quarter is the base.
Equation 2:
The Impact of the MW on Wage and Priee-Inflation; The
Wage-lnflat1on Equation.
This seetlan deals with inflation and other macroeconomic aspects
of the MW legislation.
Specifically, we analyze the potential
inflationary impact of increases in the MW and the effect of such
increases on the wages of employees and conseQuently on their labor
force status.
Brigitte Sellekaerts (1981) measures the direct and indirect wage
and price inflation impact of the MW, with particular emphasis on wage
inflation.
Sellekaerts uses the framework of a quarterly econometric
model that captures the economic interactions relevant to d1sentangle
the aggregate direct and indirect impact of the MW 1ncreases on wage
and price inflation.
35
The effect of the MW changes on wage-1nflat1on. can be eas11y
.
understood if the process 15 v;ewed as taking place 1n several
stages.
Sta9ù:
In response to higher wages and therefore higher labor
(osts, finms attempt to raise their product priees and require
employees to increase production in the short-run.
Stage 2:
A Rwage-comparison" or "ripple effett" may Dceur.
This
may be due to a quick upward adjustment in the hourly wage payment of
workers who already were making more than the new MW level prior to
iB enactment.
Such a "ripple effett" may be caused by specifie labor
contract clauses contingent upon the MW.
Stage 3:
A direct increase Dccurs also in the hourly earnings of
workers who were previously paid less than the new MW.
stage 4:
As businesses are faced with a given labor/capital
ratio. they adjust the level and the mix of their inputs demand.
The
mix cons1sts of low-skilled labor, high-skilled labor, capital goods
and raw rnaterials used in the production process.
To m1nimize costs, the new input combination rnay involve an
increase in the use of capital, a reduction in low~skilled workers and
an increase in high-skilled labor.
Stage 5:
The new employment eQuilibriurn level, combined with the
new workers' earninqs produces a new income level, a new aggregate
demand, and after sorne adjustment, affects the productlon level.
Stage 6:
F1nally, the inflation and unemployment rates,
consistent with the new equilibrium levels of incorne, output. costs,
3&
demand for goods and factor demand and supply may in turn. in time.
raise the AHE v'a the ·spi"
over effect. M
The figure on the next page presents the transmission mechanism
of the MW effect to wage/price inflation.
•
31
unit
labor
/
costs
J
~
wages of
priee
other workers
2
4
adjustments
r
1
minimum
wages of
employmf!nt
wage
1-----
minimum
1
adjultments
5
inc:r8ase
wage workerl
~
prod uc. tivi ty
6
adjultmenh
new inc.orne
ond
7
production
18vell
Figure 4:
Transmission of the MW Effects to Wage/Price Inflation
38
Measurement of the Inflation Impact of the Minimum Wage in Previous
Studies
A number of past studies have exam;ned the inflation impact of
the MW.
Sellekaerts (1980) presents one of the mast notable efforts
made to Quantify the average size of the indirect economic sp1110ver
effect of a given MW change on wage inflation.
Table 4 displays selected existing studies of the impact of the
MW on wage inflation.
Total Impact
Direct
on Wage (W) or
Authorl
Methodology
Impact
Priee (P)
Year
%
%
MPS14
Wage detenminations
.125
.15 (P)
1975
E. Gramlich
Wage determination model
.28
.28 (W)
1976
P. Fortin
Industry wage bills
.4
.6 (P)
1978
U.S. Dol.
Wage bi 11
.37
.37 (W)
1979
~e11ekaerts Wage detenm1nation
.26
.65 (W)
980
relations
Table 4:
Comparison of the Direct and Total Inflation Impact of a 10%
Increase in the level of the MW in Various Studies
Source:
8. Sellekaerts. MW Study Commission. Vol. VI, p. 4.
39
Two lessons can be learned from these past works.
First, they
generally foeus on one or a few of the several steps in the
transmission mechanism sUmmdrized in Figure 4.
The consequence is
that wage comparision and other spill-over effects may not have been
properly evaluated.
Second, in cases where efforts were made to
consider all steps, MPS (1971) Gramlich (197b), the impact of the MW
was not traced out to capture the spill-over effects on priee
interactions and product;vity effect.
The typical estimated relationship is of the following farm:
,IP
,
Zl
J
t -J.
Where;
Wt is the percent change in the average hourly earn~nqs.
u is the unemployment rate.
MW is the percent change in the MW rate.
P represents the percent change in the consumer priee index (CP!).
and
Z denotes other variables (e.g., product;vity).
Interpretation of Results:
From Table 2, one can read that a 10%
r;se in the MW wou1d entail a direct wage effect of .4% for Fortin
(1978) and .28% for Grdm1ich and a .6% effect on Priee and .28% on
wage as total impact, respective1y.
These resu1ts must be interpreted
with caution, however.
Severa1 methodological shortcom1ngs must be
addressed if the MW impacts are assessed by waqe determ1nation
mode1s.
(a) The AHE (average hourly earn;ngs) are a weighted average
of both the MW of workers and those of other workers.
As a
consequence, econometric relationsh1ps that relate changes in AHE as a
~o
dependent variable to changes in the MW as an 1ndependent variable are
subject to the fallacy of composition, which relies on a ·part a to
expla1n the ·whole.'
Accordingly biased and therefore unreliable
estlmate of the HW effect on AHE will be obtained.
(b) Collinearity
exists between the variable that represents inflation and the MW.
(c) Finally, the other explanatory variables. unemployment variable,
productlvity and lagged consumer priee index are subject to
simultaneity biases since they are affected by changes in the MW.
The severity of the shortcomings of the simple wage determlnation
model outl;ned above led Sellerkaerts (1981) to propose a new approach
to break the multi-collinearity problem.
The Wage-Inflation Equation:
The aggregate short-run wage
inflation effect of the MW.
consistent with the theoretical
foundation of aggregate wage determination, the basic estimating
eQuation can be expressed as:
Wt ~ QO + Ql(
l
) + Q2 PCKIHW + t.Qi PC CPIt-i + Q4 GNP + ~.
TNUR
l
Definition of Variables
W , the measure of wage inflation is defined as the annual
t
percentage rate of increase in some composite measure of hourly
earnings ln the economy.
TNUR ;s the teenage unemployment rate.
[t 1s used to define the
unemployment variable because we are interested mainly in the MW
impact on youths' labor force status.
TNUR is i.nversely related to
the wage-inflation variable.
41
PCKIMW is the percent change in the Kaitz index of the minimum
wage.
PC CPI is the percent change in the consumer priee index.
GNP. the gross national project. describes the level of economic
act1v1ty.
Under the constraint that the sum of the coefficients ai is
unit y, we can avoid the multicollinearity problem between the wage
variable and priee.
By sa doing. the effett of the MW changes in
other than the impact Quarter is no longer insiqnificant; therefore we
expect the inclusion of wage-inflation equation to affect the MW
coefficient on employment/unemployment. with the magnitude and the
direction to be determined empirically.
Equation 3:
The Female labor Supply
The analysis of labor supply has an important bearing on a wide
variety of issues of econornic and social policy.
Controversies about
unemployment, especially youth unernployrnent. waqe rigidity and other
macroeconornic problems often raise Questions the study of labor supply
could provide sorne answers for.
In the past two decades or so, this
has resulted in an enonmous body of theoretical and ernpirical work.
including the emerging strength of female labor supply.
The aim of this section is to examine the major determinants of
the labor supply function of wornen, especially of married women.
5uch
a function can be included in the estimating model and tested in order
to investigate how an increase in the female labor supply can affect
the teenage labor force status.
42
A Review of the Literature on the Female Labor Supply
Much of the theory of female labor force participation is
attributable to Jacob Mincer (19&2) and 1t5 application to married
women is due to Glen Cain (19&6).
As Mincer (op. cit.) observed. the
most important phenomena in the American labor force is the secular
trend in the Labor Force Participation Rate (LFPR) of married women.
Indeed, over the past 30 years, there has been a tremendous increase
in the number of women enterinq the labor force.
Several factors
explain such a trend.
First is a pure ecanomie factor in the form of financial
pressure.
The consumption level has increased in the U.S. with a
concomitant rise in the break-even level of income.
Ta fil1 the qap
between actual income and desired consumption, wives are forced into
the marketplace to complement their husband's income by a second
salary.
The Second factor concerns the increase in the education level of
women.
Since more and more women are attending college, there are
more women skilled for primary jobs.
The possibility of promotion and
achievement that characterizes entry level jobs attract a qreat number
of women to remain in the labor force for a longer period and take
less time off for maternity.
Third is the shift from heavy and dirty industrial jobs to more
service industry.
These changes in the job structure from
manufacturinq toward clerical and technical (i.e., expanded
white-collar and professional job opportunity) are also a major reason
for women ta stay in the labor force.
43
The fourth factor is the availab11ity of time-sav1ng home
technology.
Appliances have enabled women to spend less t1me on
traditional work in the house and consequently have given them more
t1me for the marketplace.
The fifth reason concerns recent social changes 5uch as
anti-discrimination laws. affirmative action, women's liberation
movement for sexual equality. which have contributed to the
acceptability of women working in the marketplace.
Finally, the development of child care services has reduced the
time spent in child rearing.
In addition, the decline in the male
labor force participation rate has resulted in an increase in the lFPR
of women.
The male LFPR decline ;5 primarily due to early retirement
(case of older men) and more years of schooling (younger men).
Because of such trends, it is important to fonmulate the labor
supply not just for an individual in isolation, but for a household.
15
Oifferent studies on the family labor supply have been conducted.
After Mincer's (op. cit.} seminal work which provided the theoretical
framework, several empirical studies took place.
These studies are
based mainly on the cross sect10nal data from the 1960 census, and
dealt with the lFPR of Black and/or White women and have focused
primarily on married women.
Glen Cain (1966) analyzed the labor force behavior of white and
non-white married women.
William Bowen and Aldrich Finegan (1969)
examined the labor force behavior of black and white women, especially
married women with regard to different responses to specified
variables.
James Sweet (1973) focused on the employment behavior of
44
black and non-black w1th respect to variables such as family
composition and an 1nd1cator of f~nanc1al pressure (husband's incarne).
Early resu1ts by Clarence Long (1958) trom cross-sectional
analys1s by city showed that the higher the earnings of the husband,
the lower the LFPR of the married women.
The result was not found to
be very meaningful because the wife's wage effect on her participation
was not controlled for.
For that reason. such a study is termed a
1b
"ma l
" " 1
e chaUVlnlstlC 1 madl
e .
In f aet. the
"
W1 e f
s
' " "
partlCipatlon
seems to be positively related to her own wage rate.
lt;5 thus
necessary to include both the wage of the husband and that of the
wHe.
Such a model may be called the "Family Utility-Family Budget
constraint" model.
Here the utility that is maximized is total fami1y
utility, assumed to depend on total famlly consumption, on the leisure
time of each fam11y member, and on
other socio-economic variables
like the number of chi1dren and the education leve1; so that utility
is maximized subject to a fami1y budget constraint.
This total fami1y utility mode1, which was first developed by
Kosters (19&&) and has proved to be by far the most popular treatment
of family labor supply behavior.
To show just how the family labor
supply model works, suppose that all family members' wage rate rises
eQui-proportional1y because of the mandatory minimum wage. '7
Furthermore, if we assume that the prices of all consumer goods stay
the same, one may invoke Hicks' composite good theorem and treat the
aggregate of the family members' leisure time and the aqgregate of the
family consumption expenditure as two composite goods--we may label l
and C respectively.
Then when the waqe rate rises as the result of
45
the impos,t1on of a MW for each member of the fam11y. the incarne
effect tends to increase L following the wage 1ncrease.
An incarne
"compensated" equi-proportionate rise in all members' wage rate would
always reduce the composite Land increase the composite C.
Moreover.
because thls change 1n waqe rate increases consumption spendinq if we
assume normal goods. total family earnings must increase to account
for the substitution effect of the ri se in the wage; hence the need
for the other member (wife) to enter the labor force.
This model broadens the simple individual labor supply analysis
in many respects.
Flrst there is the substitution effect on the
family member's labor supply of an increase in that fam11y member's
own wage:
this is the well-known own-substitution effect.
Second,
there is the effect on the family mernber's labor supply of an incorne
compensated rise in the waqe of the other family members:
the
cross-substituUon effect.
Such an effect is positive or negative
dependinq on whether the leisure times of the family members are
complement or substitute.
Regardless of the siqn of these
cross-substitution effects, the structure of the model is such that
they will always be equal.
As Ashenfelter and Heckman (op. c1t., p.
75) put H. the rnodel implies that "an incorne compensated change in
the husband's wage rate has the same effect on the wife's work effort
as an incorne cornpensated chanqe in the wife's wage rate has on the
husband's work effort."
Finally, if the cross-substitution effects
are zero for all farnily members the only effect on one member's labor
suppl Y of arise în another mernber's waqe ;s a pure incorne effect.
At
4&
the outset, if the outcome of the own-subst1tut1on effect.
cross-substitution effect and incame effect are such that the wife is
attracted into the marketplace. she still faces the problem of gett1ng
a job.
In addition, if we assume that only low-wage. law-skill jobs
are available, she may compete with a special subgroup of jobseekers:
TEENAGERS, seeking jobs of similar characteristics, because of lack of
experience and adequate 5k;115.
Ta evaluate how the women's labor suppl y would affect teenage
unemployment (and vice versa), we suggest a variant of Mincer's
specification.
Mincer's model includes a measure of marrled women's
labor suppl y as the dependent variable and a hast of independent
variables as presented below.
1 FLS 0 Co + Cl KIMW + C2 THUR + C3 [DUC + C4 INUSB + Cs CNLD + U
Definition of Variables
Dependent Variable:
FLS. the female labor suppl y, is defined as the ratio of employed
females, 20 years of age (and older) to the female civilian labor
force 16 years of age and older.
JOdependent Variables:
KIMW is Kaitz index of the MW; it serves as a proxy for women's
reservation wage.
TNUR indicates the tee nage unemployment rate.
The major emphasis
of this study is on the relationship between the tee nage labor force
and the MW, i.e., how the MW at the outset affects youths ' labor
47
status. control11ng for relevant supply variables such as FLS.
It 15
thus important to incorporate the tee nage unemployment rate 1n the FLS
specification to adequately assess how and by how much such
relationships will influence the MW coefficient in the basic
1B
equation.
[DUC is the education level of married women (completed high
school or better).
IHUSB represents the median incarne of the husband; lt serves as a
measure of financial pressure on the wife to work in the marketplace.
The proposition that financial pressure is related to the LFPR of
women is supported by the empirical findings of Mincer (op. cit.).
Indeed, in a nationwide sample survey19 of women with respect to
factors influencing their deeision to participate or to withdraw from
the labor force, economie necessity is found to be the reason most
frequently given as the dominant motive for entering the labor force.
CHlD indicates the number of eh11dren under 6 years of age.
From a theoretieal viewpoint, we expeet FlS to be positively
related to the MW variable, ta education and positively correlated to
tee nage unemployment; FlS is expeeted on the eontrary ta be negatively
related ta the husband's incorne and ta the number of children.
Formally, we have:
.FLS
.FLS
~FlS
> 0
> 0
< 0
aKIMW
'Eoue
.IHUSB
.FLS
>
aFL~
0
> 0
aTNUR
.eHLO
But as Michael KeelyZO points out, pure theory rarely provldes
answers ta important poliey questions; only empirieal research is
48
definit1ve because many policy issues h1nge on the magnitude of the
effects of changes in the MW rates on Teenage labor force status
regard1ng the impact of the MW on Teenage employment/unemployment.
Therefore. we now turn ta the empirical estimation.
49
Endnotes
l·Teenagers· means ind1viduals 16-19 years old; in what
follows, -teenagers· and "youths-; are used interchangeably.
2Effect1ve minimum wage is a wage above what wou'd have
prevalled in the absence of 5uth legislation.
4Arnold ~atl:
"Teenage Employment Effect of State Minimum
Wage,· Journal of Huma" Resources 8 (Spring 1973), 250-256.
SAlan Fisher:
-The minimum wage and teenage unemployment:
A
corrment on the literature,· Western Economie Journal 11 (Dec. 1975),
514-524.
6Thls is 50 because of the new way of countinq the unemployed
(Surrmers Lawrence (1981).
7Minimum Wage Study Conmission. Vol. l, Chapt. 2, p. 3l.
BThis type of model is presented by Adie (1971, 1913).
9Th1s model is heavily based on James Ragan's article:
IIThe
Theoretical Ambiguity of a Minimum Wage."
Atlantic Economid Journal.
Vol. V., No. l, March, 1917.
lORagan cal1ed this a general-equilibrium effect; we believe
however that the term "sp 1l10ver effect" is more appropriate since we
are dealing here with interactions of segments of the labor market
(covered and uncovered), not with the full interaction between all
markets of the economy as a whole.
l'specification of this functional fonm in no way alters the
findings of this model.
The sole reason for specifying PB as in
equation (7) is that 1t reduces the complexity of various expressions;
the term ~ in the denominator guarantees that PB assumes a value
between 0 and 1.
12The values of awu. aPB; and aP can be derived by
'MW
,MW
'MW
total1y differentiating equations (3) (5) and (B) and using Cramer's
rule.
Since we are primarily concerned with the impact of a change of
the HW on Employment. we have set other differentiations as ide.
13As Ragan (1977) observed, the fraction of the teenage
population covered is superior, at least conceptually.
The reason is
that the impact of changes in coverage depends on where they occur.
A
change in teenage-1ntens've 1ndustry. such as retail trade, should
have a larger impact on teenage employment than a change in
50
adult-intensive 1ndustry such as manufactur1ng.
The actual caverage
teenage 15 not known; the caverage variable of this study ;5 based on
est1mated teenage [average by Brown et al. (1981).
12MPS 15 the "11- Penn- Social Science Research Coune11; 1t
uses a Quarterly econometric model of the U.S. Economy.
13For a more complete mathemat1cal derivation of the family
labor supply, see Ashenfelter and Heckman:
The Estlmation of Incarne
and Substitution Effects 1" a Model of Family Labor Supply.
Econometrica. Vol. 42. No. 1. Jan. 1974.
16For an example of the use of the male chauvinistic model, see
Bowen and Finegan (1965, 1966, 1969).
171t 15 assumed that all family members work in the covered
sectar.
18That is equation 1 developed earl;er.
19Carl Rosenfelle and Vera Pervella:
"Why Women Start and Stop
Work.ing:
A Study in Mobl11ty.l'
Monthly labor Review, Sept. 1965,
p. 1077.
20H1chael Keely:
labor Supply and Public Policy; a Critical
Review.
Academie Press. 1981.
51
CHAPTER III
SAHPlE DATA SOURCES AND EHPIRICAl FINDINGS
This chapter examines the impact of the minimum wage on Teenage
Labor Force Status. controlling simultaneously for the female labor
supply (FlS) and wage-inflation (WI).
Sample Data Sources
As did most previous studies listed in Table l (page 7), the data
are collected from monthly series of the Current Population Survey
(CPS).
The data for the three basic equat10ns of tee nage emp l oyment ,
teenage unemployment and teenage labor force participation are
obtained mainly From the CPS.'
Wage-1nflat1on and fernale 'abor supply data were partially drawn
From the CPS and part1ally From the Cit1bank data base called
2
C1t1ba'e.
Ta make our results comparable w1th previous ones. we used
quarterly data (Quarterly averages of the monthly observations).
The
per;od of study is from the first quarter of 1954 to the fourth
Quarter of 1959.
List of Variables
A major concern in previous empir;cal works was to provide an
answer to the questions about what definition of the labor force
status to use as a dependent variable.
In order to test the
52
perfonmance of alternative definitions of the labor force status. we
analyze the est1mat'ng equat10ns us1ng the three defin1tions of the
labor force status of teenagers:
Teenage employment (TEMP). teenage
unemployment rate (THUR) and teenage labor force participation (TLFP)
respectively.
A final set of equations makes the estimation using s'multaneous
equat10ns with the two-stage least squares technique (2SLS).
Table 5 presents the 115t of all variables utilized in all the
eguatlons.
Table 5:
List of Variables.
Symbols
Definitions
1 - TEMP:
Teenage employment; ratio of civilian employment
to the c;vilian population for teenagers (1&-19).
2 - HW:
Minimum wage index:
the Kaitz index of the
min1mum wage; it is the product of the relative
level of the minimum wage (compared with average
hourly earn1ngs) times the fraction of teenagers
subject to the min1mum wage provision (coverage).
3 - PRIMEUR:
Prime age employment rate for males aged 24-54.
4 - SV:
The fraction of the teenagers aged 16-19 who are
actual1y 16-17.
5. - AFP:
The ratio of teenagers in the anmed forces to the
total teenage population.
6 - EFTPP:
The ratio of enrollments in federal training and
employment programs for youths 16-21 years old.
7 - POP:
The ratio of the teenage civilian population to
the total civilian population.
8 - EOP:
The ratio of teenagers enrolled in school to
teenage clvi11an population.
53
9-Pcwel:
Pr1ce-deflated welfare variable (A1d to Fam111es
w1th Dependent Ch11dren; Food Stamp programs).
10 - 1:
A l1near t1me trend.
11 - T50:
lime sQuared (to capture technolog1cal change).
12 - 02 03 04
Dummy variables for the second. th1rd and fourth
Quarters.
13 - FLS:
Female labor supply:
ratio of employed females 20
years of age and older to the c;vilian labor force.
14 - WI:
Measure of wage-1nflation:
annual percentage rate
of 1ncrease in the average hourly earning.
15 - THUR
Teenage unemployment rate:
ratio of civilian
unemployment to civilian labor force for teenagers.
16 - TLFP
Teenage labor force participation rate:
ratio of
c;v11ian labor force to civilian population for
teenagers.
17 - PCHW
The percent change in the Kaitz index of the
minimum wage.
lB - PIHFL
The price inflation rate:
the percent change in
the consumer price index.
19 - GHP
Gross Hat10nal Product:
measure of economic
perfonnance.
20 - [DUC:3
The median years of school completed. 4
21 - IHusb 5
Median incorne of husband (quarterly).6
22 - CHLD
Humber of children under 6 years of age. 7
23
LBOUTU
Labor Product1v1ty.
54
Emp1r1cal F1ndlngs
The f1nd1ngs are organ1zed in three major parts, each part using
a new def1n1tion of labor force status:
employment. unemployment and
labor force participation respectively.
Each part con51st5 of three sections:
Section l is a repl1cation of Professar Brown et al. single
equatlon model.
Section 2 extends the Brown et al. model by 1ntroducing new
control variables, female labor supply and wage-inflation as
exogeneous variables.
Section 3 presents a simultaneous equations analysis, with the
two-stage least squares technique (2SlS) with the inclusion of flS and
Wl as endogenous variables.
Part 1:
Employment Equation
Section 1:
Replication of the Professar Brown et al. Single Equation
Model
ln this section we examine the effect of an 1ncrease in the
minimum wage on teenage employment.
Different specif1cations and
funct10nal foms are cons1dered, following the Brown et al. framework.
The basic equation 1s:
TEMP
00 • B1MW • 0ZPRIMEUR • 03SY • 04AFP • 05EFTPP • O~POP •
07 T • 0B TSQ • 090Z • 01003 • 01104 • e.
55
Table 6:
Estimated impact of an 1ncrease in the MW on teenage
employment:
Basic eguation OlSB l1near.
EMp· .794 - 0.14089MW - 0.54865Y
+ 0.0728EFTPP - 0.8112POP +
(3.08)*
(5.68)
(.256)
(1.36)
0.611AFP - 1.31 PRIMEUR - 0.000921 + 0.0000205150 + 0.037102 +
(2.036)
(9.57)
(1. 74)
(5.44)
(9.885)
0.1008603 + 0.0158904
(24.12)
(4.020)
R2 • . 95
R2 • . 94
F(11; 92) = 176.46
ow = .8125
(* t-stat1stics are in parentheses.)
We can glean from Table 6 above that the coefficient of the
minimum wage is negative (-0.140B) and statistically significant
(t-va1ue = 2.56).
Based on such f1nd1ngs, previous research concluded that the MW
is the detenminant of the reduction in the teenage employment.
However, the very low Durb1n-Watson stat1stics (OW = .B25) shows
considerable serial correlation and hence confers little mean1ng to
such results.
The results of different specifications of the basic
employment eQuation are tabulated in Appendix A.
Each specification
depends upon which control variables are included or excluded from the
above basic eQuat1on.
To purge such serial correlation From the result. we estimated
9
the GlS
equation in arder to test how robust the OLS results are.
56
Table 1:
Estimated impact of an increase 1n the MW on tee nage
employment:
Bas1c eguat10ni GlS linear.
TEHP = 0.7304 - 0.06307MW - 1.1191PRIMEUR - 0.509654 - 0.567EFTPP -
(0.990)
(5.76)
(2.944)
(3.21)
0.50BOPOP + 0.465AFP - 0.001147 + 0.0000212750 + 0.04002 +
(.373)
(1.01)
(0.799)
(2.20)
(14.13)
0.107 9503 + 0.019704
(30.26)
(6.16)
R2 = .956
R2 = 0.953
F(ll; 91) = 331.04
ow = 1.922
RHO = • JOB
t(RHO) = 10.19
From Table 7, we observe three major changes.
First. the
estimated coefficient of the MW is reduced from -0.140S to -0.0&307,
which indicates an upward bias in the OlS estimates of about 0.017S.
This implies that failure to correct for the serial correlation yields
biased estlmates.
Second. the GlS results reveal that the MW
coefficient is no longer statlst1cally signlficant (t = 0.99) while lt
was signiflcant under OLS (t = 3.0S).
Third. the two eQuations seem
to perform well; both coefficients of detenmination adjusted for the
degree of freedom are high (0.9&).
Still. in the 1nterpretation of the results, greater emphasis
should be put on the GlS estimates.
Appendix B displays the
regression results of the different specificatlons under GlS.
Because the regresslon coefficients are sensitive to the units of
measurement, lt is dlfflcult to compare with accuracy different
estimates.
To solve such a problem, the coefficient of the policy
variable has been converted into elast1cities and presented ln Table S.
51
Table 8:
Elastic1ties:
The estimated effect of a 10 percent increase
in the MW on teenage employment.
OLS
6LS
Basic equation
li nea r
11 near
-1.01
-.45
EHoct of MW
(3.0B)
( .99)
Table 8 shows that a 10% increase in the MW yields a 1.02%
reduction in teenage employment under OLS linear and only 0.45%
reduction under &LS.
5Uth a differential 1n the estimates could lead
to an inaccurate analysis if one relies on OLS results as d1d many
preyious studies.
l1ke Professor Brown et al., we tested the basic equatlon under
severa' specifications in conjunction with different functional
forms.
The various specifications depend upon which control variables
is included.
The functional fonms used are OLS 11near, GLS linear. OLS
loqarithmic and GlS logarithm;c,lO
Table 9:
Estimated 1mpact of an 1ncrease in the MW on teenage
employmentj basic eguation OLS logar1thm;c.
lnTEMP = aO + allnMW + a2lnPRIMEUR + a3lnPOP + a4SY + aSAFP +
"6EFTPP + "1 T + "BTSO + "9Q2 + "10Q3 + "1104 + y
1nlEMP = 1.699 - 0.106091nMW - .11191nPRIMEUR - 0.31931nPOP - 1. 30BSY
(3.14)
(9.52)
(2.0B)
(5.3B)
0.1065AFP - 0.1135EFTPP - 0.00093T + 0.0000409TSQ + 0.0945Q2 +
( .B60)
( .159)
(0.610)
(3.Bl1)
(9.9B)
.234403 + 0.043B04
11.43)
(4.45)
2
1
R = 0.941
R = 0.941
F(11: 91) = 151.43
DW = 0.90B
58
The change \\n the funct10nal form of the basic employment
equat10n (OLS logar1thm1c) dld not bring much change 1n the results as
shown 1n Table 9.
A 10 percent 1ncrease in the MW would yield a 1.0&
percent reduction 1n teenage employment, close to the OLS linear form
(1.02%).
The basic stat1st1cs seem about 1dentical, especially the
very low Ourbin-Watson stat1stics. indicat1ng again the presence of
serial correlation.
The complete regression results of the OLS logarithmic function
combined with the different specifications are presented 1n Appendix C.
Based on the OLS logar1thmic results. one concludes that a 10
percent increase in the MW leads to a reduction in teenage employment
of 1.0& percent.
But because of the very 10w Durbin-Watson
statist1cs. we will resort aga;n to the GlS technique.
Table 10:
Estimated impact of an increase in the MW on teenage
employment:
Basic eguat10n GlS logarithmic.
InTEHP = -1.97& - 0.08841nMW - 0.11811nPRIHEUR - 0.43&lnPOP - 1.2925Y
(1.918)
(&.0&)
(1.33)
(3.07&)
.524AFP
1.801EFTPP + 0.00078&T + 0.0000321T50 + 0.09&802 +
(.427)
(3.74)
(.218)
(1.32)
(12.33)
.24503 + 0.04&304
(24.90)
(5.33)
R2 = .9&7
R2 = .9&3
F(11; 91) = 239.89
OW = 2.021
RHO = .&47
(8.&3)
59
Compared to the OLS logar1thm1c. the GLS produces a decl1ne 1n
the coefficient of the MW.
A 10 percent 1ncrease in the HW will now
lead to a 0.88% eut in teenage employment as opposed to 1.061 in the
OLS logarithmic.
The t-value has declined and the D-W has improved.
ln the GLS logar1thm1c. the negative impact of the HW on teenage
employment is minimal (-0.088).
The full regression results of the
GLS logarithmic 1n conjunction with the different specifications are
displayed in Appendix D.
ln Table 11 on the next page. we present the results in
elasticities of a 10 percent ;ncrease in the MW on teenage employment,
using the various specifications and functional fonms concom1tantly.
Except for the case of the basic eQuat'on with the welfare
variable Pcwel. in the logarithmic fonm LPcwel, all other effects of
all different specifications have negl1gible impact.
The welfare
variable was added to the basic equation mainly to test whether an
expansion in welfare benef1ts to eligible teenagers was a significant
deterrent to employment.
lt affects the logarithmic equations but has
~irtually no effect in the linear ~ersion.
Likewise. the addition of [FTP?
the enrollment/populat1on ratio
in federal training programs. reduces the estimated HW effect in OLS
equations, but has l1ttle effect 1n the GLS equations.
O~erall. no noticeable change in the MW coefficient emerged.
A
10 percent increase in the MW still would yield about 0.68% to 1.48%
reduction in teenage employment.
How would such estimates react to new relevant control variables?
Table 11:
Ela~ticit1es:
E~t1mated effect of a 10 percent 1ncrea~e in the m1ntmum wage
on teenage employment:
var10u~ ~pec1f1cat10n~ and funct10nal fonms.
pec1f1ca-
Effect of
OLS
GlS
OLS
GlS
tions
the
L1near
L1 near
Loaarithmic
Looarithmic
. Basic
MW
-1.0205
-0.454
-1 .060
-0.884
(3.08)
(0.9908)
(3.249)
(1.918)
~. 8asic - TSQ
MW
-0.0965
-0.4B
- 1 .05B
-0.931
(2.5601)
( 1. Dl )
(3.022)
(1.9B)
~. Basic - SV
MW
-0.071BB
-0.2717
-0.794
-0.6B8
(1.909)
(.555)
2.1 5B)
(1.377)
. Ba~ic - AFP
MW
-0.OB16
-0.3574
-0.965
-0.B37
(2.549)
( .7B7)
(3.147)
(1.B73)
5. Basle - EFTPP
MW
-O.1017B
-0.5139
-1.059
-0.91
(3.1025)
(1.123)
(3.26)
(2.03)
6. Ba~tc - POP
MW
-0.1031
-0.4371
-1.126
-0.84
(3.115)
(.957)
(3.403)
(1. BO)
7. Ba~ic
MW
-1.1113
-0.5584
-1.599
-1.469
+ Pcwel
(3.89)
(1.30)
(5.957)
(4.33)
B. Ba~ic
MW
-0.752
-0.378
-0.B25
-0.B21
+ EOP
(2.1006)
(0.B28)
(2.50B)
(1.819)
0'>
o
&1
Section 2:
Extension of the Basic Model:
Introduction of Female
labor Supp1y (FlS) and Wage-lnf1at1on (WI).
The most fundamental change in the American labor force is the
tremendous increase in the female labor supply.
In this section, we
introduce the female labor supply variable defined as the ratio of
total employed females 20 years of age and over. to the total civilian
fema1e 1abor force 16 years of age and over.
The data were obtained
From Citibase and are all seasonally adjusted.
Wage-inflatlon 1s added to capture the effect of inflation on
teenage employment.
Table 12 below presents the results of the basic
employment eQuation with the inclusion of the female labor supply as
an exogeneous variable.
Table 13 displays the results of the basic
eQuation when the female labor supply and wage-inflation are both
introduced.
The extended basic eQuatlon 1s:
TEMP • bD + b MW + b PRIMEUR + b SY • b AFP + b EFTPP + b&POP +
1
2
3
4
5
b T
TSO
0
0
FlS
7
+ b8
+ b9 2 + b1003 + b11 4 + b12
+ S.
Table 12:
Effect of an increase in the m1nlmum waQe on teenage
employment in the presence of female labor supply:
OLS l inear
TEMP = 2.127 - 0.12&8MW - 2.372PRIMEUR - 0.48454SY + 0.200&AFP -
(3.00&)
(8.477)
(5.387)
(0.&85)
2.b27POP
0.0594EFTPP - 0.0002541 + 0.0000134TSO + 0.025802 +
(3.77)
(0.22b)
(0.493)
(5.934)
(5.934)
0.08n03 + 0.0034904 - 1.33&FlS
(17.35)
(0.748)
(4.22&)
~2 =
0.9b2
jl2 =
0.957
F(1291)= 192.90
DW = 0.942
62
Table 12 d1splays a strong negatlve impact of the female labor
supply variable on the teenage employment level; -1.336 compared w1th
the MW effect of -0.1268.
Both variables are stat1stically
sign1ficant w1th t-values of 4.22 for FlS and 3.006 for MW.
The negative s1gn of the FlS variable conf1nms previous studies
that wornen and teenagers may compete for some jobs (Hamenmesch. Grant,
1981).
The magnitude of 5uth an adverse effect is. however, stronger
than we expected.
5uth an effect is questionable because of the very
low Durb1n-Watson statistic (DW '" 0.942).
lt rnay be a spurious effect
due to the presence of plausible serial correlation, the correct~on of
which requ1res the GlS technique.
But before applying the GlS technique, we present the resu1ts of
the same basic equation in the presence of both the fema1e labor
supply and wage-inf1ation.
Table 13:
Basic eguation extented to FlS and WI; OLS linear
TEHP = 1.120 - 0.127HW - 2.357PRIHEUR - 0.4835Y + 0.199AFP - 2.588POP
(3.007)
(8.34)
(O.483)
(0.619)
(3.68)
0.0575EF7PP - 0.002137 + 0.0000133750 + 0.026302 + 0.087603 +
(O.218)
(0.527)
(3.44)
(5.90)
(17.14)
0.00396Q4 - 1.334FL5 + 0.0692WI
(0.832)
(4.20)
(0.542)
R2 = 0.962
R2 = .956 F(13 90) = 17b.10
OW = 0.958
&3
Add1ng wage-1nf1at10n brings 1ittle change to the regression
coefficients; thus the 'nterpretation wou1d be ident1cal to the one
dea11ng with Table 12.
However, wage inflation has a very small but
positive effect on TEMP wh1ch is statistically insignificant.
With
the low DW statistic, one cannot draw any sensible conclusion.
We
will therefore correct for the serial correlation using GlS linear.
Table 14:
Extented model in the presence of FlS and WI:
GLS linear
IrEMP = 1.14 - 0.02848141/ - 1. 710PRIMEUR - 0.445Y + 0.098AFP -
(0.45&)
(&.707)
(2.59)
(0.215)
0.&54EFTPP - 1.12JPOP - O.OOl&lT + 0.000021T50 + 0.033702 +
3.88)
(0.82)
(1.10)
(2.19)
(10.03
0.1003 + 0.012&04 - 1.090FL5 + 0.0088WI
(23.9&)
(3.4&)
(3.4&)
(0.121)
~2 = 0.979
F(13
89) = 287.73
OW = 1.94
RHO=0.73
(RHO) = 10.84
With the generalized least squares technique, when the female labor
supply (FlS) and wage-inflation are present. we note quite a drastic
change in the regression coefficients as displayed in Table 14.
The regression coefficient of the MW variable all but van1shes
(-O.02848) and becomes statistically insign1ficant (t-value ~ 0.456).
However. the adverse effect is still expressed by the negative s;gn of
the MW; but with the magnitude great'Iy reduced and insignif1cant. one
•
64
cannat state that the MW 1s lead1ng to a decisive reduction 1n teenage
employment.
The female labor supply variable. though. still rna1ntains a
strong negatlve impact on teenage employment (-1.090); and such impact
;5 statistically sign1f1cant (t-value ~ 3.46).
Wage inflation, however. has a negligible impact on teenage
employment.
It5 magnitude 1s small; it5 s1gn positive and
statistically insignif1cant (D.OOBS) and (0.121) respectively.
Therefore one cannat draw any meaningful conclusions as to whetner
wage-inflation affects teenage employment; although previous studies
did conclude that MW has an effect on wage inflation.
On the whole. the female labor supply appears to influence
teenage employment great1y.
It is a major factor in the reduction of
teenage employment.
For instance, if we convert the resu1ts into e1asticities, one
can say that a 10 percent increase in the female labor supp1y would
1ead to a 21.8 percent decrease in teenage emp10yment. whereas a 10
percent increase in the MW is known to reduce teenage employment by l
to 3 percent.
for this reason, one shou1d not attribute the decline
in teenage employment to the MW alone so as to suggest a subminimum
wage as a policy solution.
Moreover. it ;s well-known ;n the literature that female labor
supply ;s determined by several variables and hence must be est;mated
endogeneously.
Sim1larly, wage-inflat;on is influenced by several
variables we discussed in Chapter 2.
&5
Accord1ngly, in arder to measure the effect of the FLS and WI
with accuracy. one must introduce these two variables as endogeneous.
We will therefore turn to a simultaneous equatlons estimation in the
next section.
Section 3:
A S1multaneous Equations Analysis of the Extented Basic
Teenage Employment Hodel:
Iwo Stage least Squares Estimation.
In thls section we present the joint estimation of teenage
employment (TEHP). female labor supply (FlS) and wage-inflation (WI).
Two alternative techniques are used.
The tirst is the strict two
stage least squares (2 SLS) and the second is the instrumental
variables method.
Since all the dlfferent specifications presented in sections l
and 2 did not change the regression coefficients substantially, we
will concentrate on the extented basic eQuation.
On the other hand.
considerable changes do emerge when one uses GLS instead of OLS;
therefore we analyze the OLS-GlS gap to gauge the sensitivity of the
estimates to the correction of serial correlation obtained by the GlS
estimation.
More weight must be put on the GlS results in the
interpretat10n phase for adeQuate policy fonmulation. both in single
equat\\on estimation and in joint estimation.
We describe first the specification of the estimating eQuatlons
of the two endogeneous variables, fLS and WI.
The extended basic
employment remains the same as before.
66
Female Labor Supply Equation
Chapter 2 descr1bes the specification of the female labor suppl y
equat10n as encountered in mast prev10us stud~es.
We further add the welfare variable 1n arder to gauge to what
extent the welfare benefits create disincentives to work as found in
past research.
FLS ; Co + C MW + C lTNUR + CJEOUC + C lHUSB + C CHLO + C PCWEL + U
1
2
4
5
6
All variables are defined in Chapter 2.
However. incarne of
husband (IHUS8) is expressed as the percentage change in the median
incarne of husband, in real ter1llS.
Table 15 shows the empirical results of the FLS equation using
OLS linear.
Table 15:
Estimation of the female labor supply:
OLS linear.
FLS = 0.699 + 0.OJ02MW - 0.0006BIHUSB - O.OOJOBCHLD + 0.0162EOUC +
(1.64J)
(2.679)
(0.904)
(0.965)
0.0027lTNUR - 1.1912 PCWEL
(5.04)
(7.24 )
R2 = 0.874
ii2 = O.B64
F(6; 96)
119.7B
OW = 0.74S
The results in Table 15. conf1rm mûst prev10us studies:
the
h1gher MW index, other things being eQual, will induce more women into
the labor market.
The impact seems not too strong but statistically
slgnificant.
Moreover, income of husband has the expected negative sign and is
sign1f1cant although of small magnitude.
&7
On the other hand. the variable representing the number of
chl1dren under 6 years of age has almost no notable effect on the
female labor supply.
Although 1t does have an adverse effett, it is
of little significance bath ln magnitude and statistical1y.
This may
reveal the growing availability of chl1d tare services.
Furthenmore,
thls pattern perhaps reflects the faet that women now have fewer
chl1dren on average.
ConseQuently the number of children appears not
to be as serious a hindrance to employment as before.
Education has a positive influence on female labor supply.
This
implies that the higher the education level of women. the more likely
they are to be in the job market.
Likewise, the teenage unemployment rate 1s pos1tively related to
FLS.
This supports Hamenmesh and Grant (1981) findings that women and
teenagers may be close substitutes for sorne jobs.
However, the most notable result is the statistically significant
and strong negative impact of the welfare variable on FLS.
The
results demonstrate that welfare benef1ts are a deterrent to women to
enter the labor market, ceteris paribus.
Overall, the signs of the FLS eQuation estimates reflect the
expected behavior of the explanatory variables.
Wage-Inflat1on Equation
The second endogeneous variable we introduce ;s the
wage-inflation.
In Chapter 2, we analyzed the potential inflationary
impact of increases in the minimum wage and what the effect such
68
increases might have on wages of employees; and consequently how this
will affect their employment level. espec1ally for youths.
The following equat10n is a modified vers10n of Sellekaerts wage
inflation equat10n.
WI • 90 + 91PCMW + 92ITNUR + 93LBOUTU + 94PINFL + 95GNP + U
where PCMW is the percentage change in the minimum wage and PINFl is
the percentage change in the consumer price index.
The next table puts forth the regression results.
Table lb:
Regression coefficients of the wage-inflation eguation:
OLS linear
WI = 0.0018 + 0.0297PCMW + 0.577PINFL + 0.000107LBOUTU + 0.00025ITNUR
(1.9)
(2.95)
(0.80)
(0.318)
0.115GNP
(0.107
R2 • 0.314
R2 = 0.279; F(5
97) = 8.91
OW = 2.52
We observe that all the five explanatory variables have the
anticipated positive impact on wage-inflation but only the percentage
change in the MW and the percentage change in price inflation are
statistically significant.
How and to what extent will the endogeneously estimated values of
wage-inflat10n and the female labor supply affect the results of the
basic extended model?
The answer resides in the simultaneous
estimation to wh1ch we now turn.
69
Two-Stage least Squares (2 SlS) Estimation of the Teenage Employment
Equation.
ln the 2 SlS procedure, we estimate the reduced-form equat10ns by
OLS.
This involves regressing FlS and Wl on their respective
explanatory variables.
We get the pred1cted value of FlS and WI.
In
the second stage, we regress TEMP on the exogeneous variables and the
predicted values of FlS and WI.
The basic 1dea in the 2 SLS is to substitute for the endogeneous
variables which are correlated with the residuals, a linear function
of all the predetermined variables.
Since these variables are
uncorrelated ln probability 11mit with the residuals, this 2 SlS
procedure will yield consistent estimates of the parameters.
Applying the above 2 SlS method, we estimated the teenage
employment equation.
The use of the 2 SlS was justified by the fa ct
11
that the system of equat10ns is overidentified.
The results are
presented in Table 17.
Table 17:
Est1mated impact of an increase in the MW. FlS and Wl on
teenage employment.
2 SlS.
EMP = 1.60 - 0.942MW - 1.155PRIMEUR - 0.53B5Y • 0.3B4AFP -
(2.15)
(B.09)
(5.73)
(1.37)
0.0474EF7PP - 1.09BPOP - 0.001247 • 0.0000172750 • 0.03602 •
(0.172)
(1.B77)
(2.32)
(4.66)
(B.B9)
0.101503 .0.016704 - 0.0902BFL5 • .2BWI
(25.49)
(4.27)
(3.167)
(3.03)
R2 • 0.963
îl2 • 0.957
F(13; B9) • 17B.B3
OW
1. 19
10
We can glean From the preced1ng table that two out of the three
policy variables. the minimum wage 1ndex and the fernale labor supply
have a sign1f1cant negative impact on the teenage employment level.
This 15 close to the single eQuation estimation results of the
extended basic model.
However, 1n the 2 SlS method. the magnitude of
the FlS 1mpact is reduced From -1.336 to -0.9025; that is a bias of
-0.434 exists between the two methods.
The other policy variable
wage-inflation, still has a positive effect on the teenage employment.
which becomes statistically significant and of a greater magnitude
than that of the OLS case.
Apparently, an increase in the wage
attracts more teenagers into the job market.
Nevertheless, in the first stage of our 2 Sl5 procedure, the OLS
estimation of the FlS equation has a very low Ourbin-Watson statistic;
a sign of the presence of seri al correlation.
Thus the above 2 SlS
method results must be interpreted with caution.
To correct for such seri al correlation, we resort again to the
GlS estimation technique.
The estima tes of the 2 SLS purged From any
seri al correlation are put forth in Table 18.
71
Table 18:
Est1mated effect of an 1ncrease 1n the minimum wage, female
labor suppl y and wage inflation teenage employment:
2 SlS-GlS.
EHP • 1.98 - 0.0484HW - 1.74PRIHEUR - 0.477SY - 0.00873AFP -
(0.77&)
(&.93)
(3.02)
(0.0198)
0.&47EFTPP - 1.073POP - 0.00195T + 0.0000228TSQ + 0.034502 +
(3.82)
(0.839)
(1.41 )
(2.50)
(10.42)
0.00003 + 0.01204 - 1. 337lFlS + 0.173WI
(24.21)
(3.18)
(4.05)
(0.44)
2 • 0.980
R2 • 0.977 F(13; 88) • 300.&8
DW· 1.89
RHO • . &9
t(RHO) • 9.84
The general1zed least squares estimation of the 2 SLS of the
extended basic model reveals sorne interesting results.
First. the
direction of the impact of all three policy variables is maintained.
An increase in the MW and an increase in the fLS bath lead to a
dec11ne in teenage employment.
Wage-1nflation on the other hand
continues to maintain a positive but insignificant effect on youth
employment.
Second, the magnitude of the impact of the female labor supply
variable ;5 greater than in the previous 2 SLS without correction for
seri al correlation.
The FLS coefficient is -1.337, close to the
estimate of the extended OLS single equation results; it is also
siqnificant (t-value = 4.05).
Third, the most striking change emerges from the fact that the
minimum wage variable is no longer statistically significant
(t-value = 0.076) and the impact all but vanishes (-0.0484).
Fourth. the wage-1nflation impact continues to be insignificant.
72
This 15 important 1n the analysis of the issue at hand.
lndeed.
the 2 SLS-GLS shows clearly that the increase in the female labor
supply 1s the major factor 1n the decline in the teenage employment;
and this decl1ne 15 not due to an increase in the minimum wage alone.
This will have important implications with respect to the issue
of establishing a subminimum wage for teenagers.
But before coming to a firm conclusion, we wfll test the robust
quality of the 2 SLS-GLS results using another method of simultaneous
equations estlmation.
Instrumental Variables Estimation Method of the Simultaneous Equations
As an alternative to the 2 SLS, we introduce the instrumental
variable method.
We argued earlier that in a regression equation
where the explanatory variable 1s correlated w1th the residual. we
cannot get consistent estimates for the parameters by using OLS.
ln a
simultaneous equat10ns model, this is no longer a problem because the
exogeneous variables not in the equations can be used as instrumental
variables.
Such a method 1S applied to our model and the results are
displayed in Table 19.
73
Table 19:
Est1mated effects of an 1ncrease in MW. FLS and Wl on
teenage emDloyment:
Instrumental variable estimation.
EHP = 3.193 - 0.114704MW - 3.184PRIHEUR - 0.4255Y - 0.1288AFP -
(2.478)
(4.044)
(4.18)
(0.36)
3.92POP - 0.16EFTPP + 0.000149T + 0.000008T50 + 0. 017802 +
(4.19)
(0.559)
(0.240)
(1. 55)
(2.83 )
0. 0771 03 - 0.0054 04 - 2.42702FlS +0.197WI
(10. Tl)
(0.808)
(4.044)
(0.377)
OW = 1.1163
An examination of Table 19 exhibits sorne features noticed
earlîer.
The MW and the FLS have adverse effects on teenaqe
employment; bath are siqnificant, but the fLS has a greater impact
than before.
Wage inflation has an insignificant and positive impact
on TEMP.
8ut the method used to produce Table 19 does not take iota
consideration any correlation in the series.
To encompass such
shortcomings. we experiment w1th the use of the instrumental variables
method with an option to correct for serial correlation.
The results
are shown in Table 20.
Table 20:
Estimated effet! of an increase in FLS. WI and HW on
teenage employment. instrumental variable
GLS.
EHP = 1.29
0.0259088MW - 1.39PRIHEUR - 0.466SY + 0.0754AFP -
(0.397)
(4.319)
(2.69)
(0.153)
0.345POP - 0.632EFTPP - 0.00228T + 0.000026TSO - 0.037702 +
0.239)
(3.68)
(1.41 )
(2.49 )
(9.27)
0. 1047 03 + 0. 016804 - 0.641569FlS + 0.0489728WI
20.88)
(3.77)
(1.34)
(0.571 )
OW
l .982
74
With the correction for serial correlat1on. we note sorne
important changes as under the 2 SlS - GlS case.
As a matter of facto the three policy variables behave almost the
same way as under the 2 SlS - GlS.
The minimum wage increase has
little impact on teenage employment, although 1t maintains its adverse
effect.
Wage-inflation aga;n has a negl1gible impact on TEH? and is
;ns;gnificant.
The female labor supply. on the ather hand, ;s statistically
sign1ficant and has a larger negative impact on teenage employment.
This supports our comments that an increase in fLS affects teenage
employment much more heavily than the minimum wage.
As a 1ast test of the consistency and robust quality of the
results. other exper1ments were undertaken using different functional
fonms. namely logarithmic funct10nal fonms.
The elasticities obtained
reflect the tendency observed earlier.
Female labor suppl y increases
seem to affect the youth employment more adverseTy than ;ncreases ;n
the minimum wage.
The two negative effects of ;ncreases in the FlS
and in the MW outweight the small positive impact of wage-inflatl0n in
such a way that teenage employment declines overall.
We summar;ze in Table 21, the results of al1 the different
exper;ment~ in elasticities.
75
Table 21:
Coefficients of elasticity;
estimated impact of a 10
percent 1ncrease in female labor supply. wage inflation and
minimum wage on teenage employmentj Summary Table.
pecif1cat1ons and/or
Elfect of
Elfect of
Elfect 0
unctional Forros
HW
FlS
WI
1 - Basic - OLS
-1.025
NI'
NI'
(3.08 \\
2 - Basic - GlS
-0.454
NI
NI
(0.99\\
3 - Basic - log OLS
-1 .060
NI
NI
13.24 \\
4 - Basic - log GLS
-0.884
NI
NI
11.918\\
5 - Extended Basic - OLS
-0.915
-26.77
+0.022
13.007\\
14.20\\
(0.542\\
6 - Extended Bas le - GlS
~?' 204
-;l.BI\\
+0.0309
0.456
3.46
10.121
7 - Extended Basic log OLS
-1.06
-39.15
+0.427
(3.90 \\
16.50\\
( .417\\
8 - Extended Basic Log GlS
-0.86
-25.67
+0.54
(2.19\\
13.75\\
1.711\\
9 - 2SlS Extended Basic
~~.6;~
-18.12
+0.098
with first staoe OLS
2.15
13.1671
13.03
0 - 2SlS Extended Basic
-?34~1
-;.&.B~
+0.0608
1
GLS
.776
4.05
10.441
1 - 2SlS Extended Bas;c
-0.46
-28.94
0.906
looarithmic
Il .39 \\
13.72\\
12.421
2 - 2SlS Extended Basic
-0.349
-26.83
+0.0608
GlS looarithmic
(0.7761
(4.05\\
10.44\\
3 - Instrumental Variables
-0.827
-48.71
+0.0692
Extended Bas i c
12.478\\
114.044\\
(0.3771
76
4 - Instrumenta l Variables
-0.186
-:2.8~\\
(~.01 ~~
Extended GLS
'0.397\\
1 .34
0.57\\
5 - Instrumental Variables
-0.759
-27.2
0.0010
Extended Basic
(1.80)
(2.38)
(0.103)
loaarithmic
HI* = Non included.
A scrut\\ny of the coefficients displayed in Table 21 reveals sorne
interesting facts.
Ta make those facts easy to understand, the
regression coefficients have been converted into elasticities.
The rows in Table 19 differ in the control variables as well as
in the functional forms.
Columns (2} (3) and (4) represent the three policy variables
under investigation.
lines l through 4 report the estimates from the
Basic Model.
The Bas;c Model includes the policy variable minimum
wage index.
In addition, it controls for the season of the year
(quarters;
Q2 Q3 Q4 with the first quarter Q, be1nq the
base).
A tlme trend and a quadratic time trend are also incorporated
in the Basic Model; llkewise cyclical factors such as the prime age
adult employment rate (PRIMEUR) and other supply s;de variables (SY,
AFP, EFTPP and POP) are present.
The coefficients of elast1city in line 1 through 4 are identical
to the ones in Table 11 (p. 60).
Therefore. we will not dwell any
further on their meaninqs; a 10 percent increase in the minimum wage
leads roughly to a one percent reduction in teenage employment.
lines
5 through B report the estimates of the Basic Madel extended ta
include the other two pol;cy var1ables:
the female labor supply (FlS)
and wage-inflation (WI).
11
An analysis of the estimated coeff1cients of elasticity of the
extended Basic Model prompts several comments.
l - The OlS-GlS senstivity persists. with the elasticities
obtained under GLS smaller than the ones obtained under OLS.
Moreover. the OLS-GLS gap persists under the l;near as well as the
logarithmic form.
2 - Perhaps the most striking finding ;5 the fact that the female
labor supply has a great negative impact on youth employment.
A 10
percent increase in the FlS yields a decrease between 21.81 percent
and 26.11 percent in the teenage employment as shown in Column (3).
Furthenmore, such a strong adverse effect ;5 consistently
statistically significant (3.46 ~ t-value < 6.50).
This impact and
significance are valid under the linear and logarithmic forms.
One ~y note also that the GlS coefficients are smaller than the
OLS.
3 - The elasticities in Column (4) show the impact of a 10
percent increase in the wage-inflation on youth employment.
An
increase in the wage-inflation has consistently a positive and
insignificant effect on teenage employrnent.
A 10 percent increase in
the WI leads from 0.022 percent to 0.54 percent increase in the
teenage employrnent level.
The logar;thrn;c form seerns to have a
s11ghtly greater elasticity though 1t is still insignificant.
4 - The minimum wage index continues to have its adverse effect
on TEMP.
However, there appears sorne ambiguity about its statistical
significance.
While the OLS estimates (linear and logarithmic) are
76
signif1cant and of great impact, the GLS (linear and logar1thm1c)
est1mates are insign1f1cant and of smaller magnitude.
Overall. a 10
percent increase in the minimum wage decreases the teenage employment
by 0.20 to 0.8& percent.
Based on the extended GlS results, one might conclude that the
adverse impact of the fernale labor supply is about 20 times as strong
as the negat;ve effect of the minimum wage.
To further test the robustness of the above findings, we analyze
the elasticity obtained using a joint estimation technique.
Lines 9
through 12 display the 2SLS estimation elasticities. in linear and
logarithmic form.
In either form. the ~W negative impact is insignificant.
The female labor suppl y on the other hand contlnues to have an
important adverse effect on the dependent variable.
Indeed, teenaçe
employment will be reduced from about 18.12 percent to 28.94 percent
as a result of a 10 percent increase in the female labor supply; such
a decline has statlstical siqnifiance as opposed to the MW effect.
One observes also that wage-inflation has a negligible but
positive effect on teenage employment.
Contrary to all previous estimations of the extended basic model,
the wage-inflation impact becomes significant under the 25L5
logarithmic (t-value ~ 2.42).
For the joint estimation as a whole, using the 25L5 method, the
findings seem to move in accordance with the single eQuation estimates
of the extended basic model.
79
[xcept for the wage-inflation which is s1gnificant under the 2SlS
logar1thmic. the minimum wage index has small negatlve impact on TEHP,
with more or less 5tat15t1ca1 significance.
The female labor supply
has a consistently greater and statistlcally significant adverse
effect upon TEHP.
F1nally. we conduct an additional experiment of the joint
esttmat10n to gauge the sensitivity of the results to the techniques
of estimation.
Lines 13 to 15 report the elasticity coefficients when the
instrumental variables techniques (hereafter INST) is used.
Here
again, one observes the preceding patterns of the effect of each
policy variable.
The HW has mixed impact.
A 10 percent increase in the MW would
y1eld from 0.18 percent to 0.82 percent decline in TEHP.
Such an
impact is significant under INST without the serial correlation
correction; but becomes insignificant when the serial correlation is
removed.
Host of all, the female labor suppl Y persists in its considerable
negative impact on TEMP.
TEHP will decrease from 12.87 percent to
48.71 percent following an increase of 10 percent in FLS.
Such an
impact is very s1gn'ficant.
Wage inflation has a very negligible effect. consistently
positive but insignificant.
80
Summary of Part 1
In the previous analys1s. one notes the ambiguity of the impact
of the MW on TEMP as far as the 5tatl5t1(a1 significance is
concerned.
Suth ambiguity is predicted by theory.
Although the MW has a negative impact in the single equation
analysis, 5uch an impact is not consistently sign1ficant in the
extended basic model and in the following joint estimation, the MW
impact al1 but vanishes.
Wage inflation a150 demonstrates sorne statistical inconslstency
in its neg11gible but positive impact on TEMP and no sound conclusion
can be drawn.
The mûst consistent impact has been the one of the femalelabor
supply variable.
Through various forms and estimation techniques. the
flS mainta1ns a considerable negative and significant impact on TE~P.
8ased on such findings one m1ght conclude that the female labor
suppl y does unambiguously have an adverse effect on teenage
employment,12
81
Part II:
UnemDloyment rate eguation.
In Part 1. we have examined the impact of increases 1n the
minimum wage, female labor supply and wage-inflation on teenage
employment.
In this second part, we follow the same scheme of analysis, only
the teenage unemployment rate is used as the dependent variable.
Section 1:
Replication of Brown et al. Basic Unemployment
Rate Equation
The unemployment eQuation is described as follows:
TNUR = e
+ e,MW + e PRIMEUR + e SY + e AFP + eSEFTPP + e POP +
û
2
3
4
6
e
Q
Q
Q
7T + ee TSQ + e9
3
4
2 + e10
+ e11
+ h.
We furnish considerable details in Part 1.
Here we present more
concisely the major results.
This includes results of the different
functional forms (linear and logarithmic) and method of estlmation
(OLS and 6LS).
Because no notable changes in the outcome arise trom the
different specifications in Part l, we will foeus on the basic
eQuation in this second part.
The tables below display sorne selected
regression results. followed by a succinct explanatory comment.
82
Table 22:
Estimated effect of an increase in the minimum wage on
the teenage unemployment rate:
8as1c OLS l1near.
NUR = -0.18& - 0.00522340MW + 1.&5PRIMEUR + 0.3455Y + 0.0405AFP +
(0.125)
(13.11)
(3.92)
(0.148)
0.5704POP - 0.158EFTPP + 0.00085T
0.000003T50 + 0.029502 +
(1.05)
(0.&15)
(1.7&)
(0.87)
(8.&1 )
0.007&03 + 0.0041104
(1.99)
(1.14)
R2 : 0.85&
F(ll; 92) = 49.99
il2 : 0.832
DW = 2.03
Table 23;
Estimated effect of an increase in the minimum wage on
the teenage unemployment rate:
Basic GLS linear.
NUR = -0.1&& - 0.00952489MW + 1.&4PRIMEUR + 0.33&5Y + 0.00398AFP +
(0.234)
(13.31)
(3.93)
(0.015)
0.3&POP - 0.192&EFTPP + 0.001108T -0.0000045T50 + 0.0302 +
(0.&5)
(0.75)
(2.203)
(1.29)
(8.71)
0.0044704
(1 .24)
R2 = 0.8&5
F(l1; 91)
51.34
il2 = 0.848
RHO = -0.028
it(RHO) : -0.293
83
Tabl. 24:
Est1mated impact of an increase in the mlnlmum wage on
the teenage unemployment rate:
lnTHUR
0.61 + 0.03441nMW + 0.03991nPRIMEUR + 0.121nPOP + 2.023SY +
(0.41)
(13.83)
(1.43)
(3.17)
0.89AFP - 0.18EFTPP + O.OOllT - 0.0000288T50 + 0. 19402 +
(0.46)
(0.31)
(1 .66)
(1.17)
(8.83)
0.048103 + 0.021104
(1.91)
(0.92)
2 = 0.81
F(l1: 92)
16.36
R2 = 0.81
OW=2.18
labl. 21:
Estimated impact of an increase in the mlnlmum wage on
the teenage unemployment rate:
Basic GLS logarithmic.
1nTHUR = -1.013 + 0.029699l1nMW + 0.3911nPRIMEUR + O.341nPOP +
(0.441)
(11.52)
(1.04)
1.929SY + 0.319AFP - 1.354EFTPP + 0.001931 - 0.0000423TSO +
3.8B)
(0.221)
(0.846)
(2.43)
(l.81)
0.2002 + 0.05 603 + 0.02 504
8.89)
(2.42)
(1 .01)
2 = 0.B91
F(ll; 91) = 69.91
2
0.B84
OW = 2.08
HO = -0.124
t(RHO) = -1.21B
The results put Forth in Tables 22 through 25 exhibit two salient
facts.
F;rst. the impact of an increase in the minimum wage on
teenage unemployment rate ;5 statistically insignificant; the t-value
is eQual to 0.125 on aVerage.
Second, the regression coefficient of
the minimum wage reflects an unexpected negative sign when the linear
B4
form is utilized under both OLS and GLS.
On the other hand. the
10garithmic form pro~1des the expected positi~e effect of an increase
in the minimum wage on the teenage unemployment rate.
Yet no
meaninqful interpretation can be made if one relies on the logarithmic
form.
This is due to the lack of precision obser~ed in the estimated
impact as reflected in the ~ery low t-statistics.
Additional
regression results of the basic teenaqe unemployment equation are
presented in Appendix E.
To test how sound are the above estimates.
we take up the investigation of the effect of changes in the minimum
wage on the teenage unemployment rate in the presence of _!emale labor
supply and wage-inflation.
Section 2:
Extension of the Basic Unemployment Equation
Inclusion of the Fema1e Labor Suppl Y and Wage-inflation
As we did in Section 1. we present selected reqression results of
the extended basic model, in a series of tables.
Following these
tables, we summarize the most salient features in an explanatory
conrnent.
Table 26:
Effect of an increase in the MW on teenage unemployment in
the presence of female labor supp1y and wage inflation:
extended model OLS linear.
NUR 0 0.43 + 0.0012B&lBMW + 1.17PRIMEUR -1- O.375SY -0.149AFP
(0.31 )
(4.244)
(4.247)
(0.521)
0.271 POP - O. nOEFTPp + O.OOlloT - 0.00000&3150 + 0.024302 +
(0.394)
(0.B55)
(2.30)
( 1 .00)
(5.50)
0.0012BQ3 - 0.00101 04 - 0.01941BFL5 + 0.00109WI
(0.250)
(0.347)
(1 .99)
(0.0133)
2
0.B02
F(13: 90) 0 43.52
R2
0.842
DW
2.051
0
85
Table 27:
Estimated effect of an increase in the ~ on teenage
unemployment rate in the presence of female labor supply
and wage inflation:
GLS linear.
NUR = 0.36 - 0.00214652HW + 1.228PRIH[UR + 0.36354 - 0.158AFP -
(0.533)
(4.49)
(4.25)
(0.57)
0.304POP - 0.254[F1PP + 0.00131 + 0.00000102150 + 0.02502 +
0.455)
(0.989)
(2.56)
(1.86)
(5.71)
0.0027703 - 0.004304 - 0.53188FL5 + 0.0215416WI
(0.55)
(0.09)
(1.13)
(0.16)
~2 = 0.811
F(13; 89) = 325.66
RHO = 0.04
~2 = 0.852
DW = 1.98
t(RHO) = 0.412
Table 28:
Estimated effect of an increase in the MW. FLS and WI on
teenage unemployment rate:
OLS logarithmic.
nTHUR = -1.28 + 0.03460441nHW + 0.3681nPRIH[UR + 0.3611nPOP +
(0.45)
(5.77)
(0.16)
2.125Y + 0.319AFP - 0.131[FTPP + 0.00631 - 0.0000351T50 + 0.18602
3.55)
(0.177)
(0.43)
(1.68)
(1.29)
(6.57)
0.031903 + 0.012004 - 0.94941nFL5 t 0.163931WI
1.11)
(0.40)
(0.55)
(0.199)
2 = 0.871
F(13: 90) = 46.86
2
0.852
DW = 2.16
86
Table 29:
Estimated effect of an 1ncrease in the MW, FlS and wt on
teenage unemployment rate:
GlS 10gar1thm1c .
nTNUR • -0.99 • 0.03026431nMW • 0.3991nPRIHEUR + O.3691nPOP +
(0.44 )
(6.80)
(0.8~)
1.93SY .0.3J3AfP - ,. 29EFTPP • 0.07896T - 0.0000433TSO •
(3. 6~)
(0.19)
(0.78)
(2.21)
(1. 72)
0. 2020 2 • 0.0~903 • 0.028~04 • 0.018~61nfLS • 0.4~8113WI
(J.l~)
(1.93)
(0.9~)
(0.011 1)
(0.~46)
R2 • 0.897
f ( 13; 89)
~2.6~
~2 • 0.882
DW· 2.08
RHO • -0.124
t( RHO) . -1. 276
Two major facts emerge from the results presented in Tables 26
through 29.
First, as we found under the basic unemployment model,
the extended model proved ta have very small and insignificant
regression coefficients for the MW variable.
Second, though the
female labor suppl y variable demonstrates statistically signlficant
results, its point estimate reveals a negatlve rather the anticipated
positive effect.
Wage-;nflation prov;des almost no discernible impact
on the teenage unemployment rate as shawn in its very small
coefficients and t-values.
As a whole, the extended model brings no substantive changes in
the results relative ta the basic model.
87
Section 3:
A simultaneous eguations analvs1s of the extended
teenage unemployment model:
Two-stage least squares and instrumental
variables methods.
ln this section, we present the joint estimation of the teenage
unemployment rate eQuation, female labor supply eQuat;on and
wage-inflation eQuation.
This estimation method will serve as a test of the robust quality
of the estimates.
We apply the 2 SLS and the instrumental variables
methods.
To be consistent with sections 1 and 2, we display selected
regression results in a series of related tables.
Succeeding these
tables is a summary of the major features of the results, put forth in
an explanatory comment.
Table 30:
Estimated impact of a joint ;ncrease in the MW. FlS and WI
on teenage unemployment:
2 SLS.
NUR
0.80 + 0.023195MW + 1.12PRIMEUR + 0.245SY + 0.0979AfT
(0.577)
(b.27)
(2.81 )
(0.377)
0.0024POP - 0.483EfTPP + 0.000931
0.0000048TSO + 0.02202 +
(0.0045)
(1 .89)
(2.005)
(1.47)
(5.84)
0.004303 - 0.0003204
1.00241 FLS - 0.550b30WI
(LIb)
(0.088)
(3.79)
(1. 399)
R2 _ 0.878
F(13; 90) = 49.81
R2
0.860
~~~
~~~~_ _~ ~ ~ _ _ l
88
Tab1.31:
Estlmat.d .ff.ct of a jolnt lncr.as. ln th. MW, FLS and NI
on teenage unemployment - 2 SlS-GlS.
THUR· 0.93 + 0.0180452MN + 1.057PRIMEUR + 0.239SY + 0.840AFP -
(0.422)
(6.65)
(2.58)
(0.304)
0.326POP - 0.404EFTPP + 0.0012T - 0.0000073TSO + 0.021702 +
(1.61)
(1.61)
(2.34)
(1.99)
(5.82)
0.003803 - 0.0002904 - 1.11479FLS - 0.488094NI
(1.02)
(0.082)
(4.01)
«1.20)
R2 = 0.868
F(13; 89) = 39.60
il2 = 0.849
DN = 1.98
RHO = 0.078
t(RHO) = 0.802
Table 32:
Estimated impact of an increase in the MW. FlS and WI on
teenage unemployment:
2 SlS logarithmic.
lnTHUR· 1.68 + 0.03l16381nMN + 0.3298831nPRIMEUR + 0.3173661nPOP +
(0.406)
(5.47)
(0.788)
0.564AFP - 1.527EFTPP + 1.919SY + 0.0065T - 0.0000384TSO + 0.7802 +
(0.28)
(0.84)
(3.27)
(1.77)
(1.48)
(7.16)
0.04303 + 0.012704 - 2.698951nFLS - 1.05712WI
(1.72)
(0.51)
(1.28)
(0.38)
R2 • 0.873
F(13; 90) = 47.68
il2 = 0.854
DN=2.10
89
Table 33:
Estimated impact of an 1ncrease in the MW, FlS and WI on
teenage unemDloyment:
2 5L5 logarlthm1c - GL5.
lnTNUR • -2.02 + 0.0231nMW + 0.3331nPRIMEUR + 0.1491nPOP + 0.127AFP -
(0.33)
(5.99 )
(0.39)
(0.0727)
2.18EFTPP + 1.825Y + 0.0088T - 0.000051T50 + 0.18702 + 0. 0513 03 +
(1.22)
(3.45)
(2.55)
(2.11)
(7.48)
(2.14)
0.01704 - 2.341561nFL5 - 1.09439WI
(0.69)
(1.22)
(0.42)
R2 • 0.895
F(13; 89) • 51.56
R2 • 0.880
DW· 2.05
RHO<-0.10
t(RHO) ·1.029
The results of the joint estimation of the teenage unemployment
eQuation, female labor supply and wage-înflation eQuations follow the
patterns encountered in the single equation analysis of the extended
basic unemployment mode',
First. the MW impact proved to be statistically insignificant as
noted in the extended basic unemployment model.
Second. the flS variable shows a significant effect on teenage
unemployment; but this impact has the wrong sign.
Th;rd the wage-inflation d1splays a mixed effect.
In the linear
form (Tables 30 and 31) an 1ncrease in the WI leads to a decline in
teenage unemployment.
In addition, the observed impact appears ta be
signHicant.
However, in the logarlthmic form, though still maintatning its
adverse effect, the WI impact is no longer statistically significant.
90
Overall. the results of the joint estimation proved to be not
sound and no mean1ngful conclusion can be drawn regard1ng the effett
of an increase in the MW. FlS and WI on the teenage unemployment rate.
Yet. as a last effort to test the robust Qua11ty of the
estimates. we carry on the joint estimation using the instrumental
variables method.
The results are presented in Appendix E.
A close scrutiny of the estimates obtained From the instrumental
variables method reveals the same salient features observed in the 2
SlS method. and there is no need to dwell on the se facts further.
As a summary of all the results of the teenaqe unemployment
eQuat1on, we put Forth a final table of all the coefficients converted
into elast1c1t1es, in order to better assess the impact of changes in
the policy variables on teenage unemployment rate.
Table 34:
Estimated impact of a 10 percent increase in the minimum
wage. female labor supply and wage inflation on teen~ge
unemployment rate (in percent).
pecifications and/or
Effect of
Effect of
Effect 0
unctional Forms
HW
FLS
WI
1 - Basic - OLS linear
-0.106
NI'
NI'
(0.121
2 - Basic - GlS linear
-0.194
NI
NI
(0.231
3 - Basic - OLS 1oga ri thmi c
-Hl. 344
NI
NI
(0.451
4-Basic- GLS 1oga ri thm1 c
-Hl.296
NI
NI
(0.441
5 - Basic + Pcwel
-0.915
NI
NI
OLS linear
(0.158)
91
6 - Basic + EDP
+1).45
HI
Hl
OLS linear
(0.501
7 - Extended Basic OLS
-0.02~
-35.26
O. Dl ~
linear
(0.311
Cl .991
(0.0131
8 - Extended Basic GlS
-0.043
-30.~3
0.020~
linear
(0.0531
(1. 731
(0.1~1
9 - Extended Basic OLS
0.34~
- 9.49
O. 1~
looarithmic
(0.45)
(0.55)
(0.10\\
0 - 2SlS
0.47
-37. 08
0.54
(0.57)
(3.791
( 1. 39)
1 - 2SlS - GlS
0.3~9
-~3.49
0.48
10.42\\
(4. Dl ,
{] .20 \\
2 - 2SLS logarithmic
0.31
-~&.98
-1.0~~
filS looarithmic
10.40\\
1 .28 \\
10.38
3 - 2SLS logarithmic - GlS
(~.2;\\
-23.41
~,l . O~~
0.33
{].22\\
0.42
4 - Instrumental Variables
0.073
-42.17
-0.133
Hethod
(0.085\\
Cl .32\\
10.285\\
5 - Instrumental Variables
-0.079
-28.95
O. 1~
Hethod - GlS
(0.961
Cl .29 \\
10.681
~ - Instrumental Variables
0.35
11.98
0.74
Hethod:
loaarithmic
10.4~1
(0.371
10.251
7 - Instrumental Variables
0.31
15.39
1. 39
Hethod:
GlS looarithmi
10.4bl
10.76\\
10.83\\
*Non 1ncluded.
To 1nterpret Table 34, consider for instance line 4 which
portrays a GLS estimate with a double logarithmic form.
The 0.296
means that a 10 percent increase in the MW 15 expected to increase the
teenage unemployment rate by 0.296 percentage point.
Similarly. line
7 portrays the OLS logarithmic estimate of the extended model.
A 10
92
percent 1ncrease in the MW. FlS and NI 15 expected to 1ncrease teenage
unemployment by 0.34; 9.49 and 0.16 percentage points respectl~ely.
But because of the lack of precis10n in those estimates as shown 1n
the low t-values, one might 1nterpret them with caution.
The
estimates of the policy variables derived From the joint estimation
display the same eQuivocal impact; those estimates are either very
small and însignificant (WI and HW) or significant and wrong signed
(FLS).
As a whole, the extended model of teenage unemployment provides
no substantial changes in the results relative to the basic teenage
unemployment estimates.
The joint estimation conveys no new pieces of information to the
single extended unemp10yment eQuation of teenagers.
This might tempt
one to Question how appropr1ate it is to use the unemployment rate to
assess the impact of changes in the po11cy variables on the 1abor
market conditions of teenagers.
As an alternative measure of the labor market conditions, we
carry on the analysis using the labor force participation as a
dependent variable.
•
93
Part Ill:
Labor Force Participation Rate Equation.
In this last part of the emp1rfcal 1nvestigation, we analyze the
effect of the minimum wage on the teenage labor force participation
rate (thereafter TlFP).
Defined as the ratio of the teenage civilfan labor force to the
teenage population, the TLFP offers the opportunlty to test how
increases in the HW. FlS and WI affect the participation of youths in
the labor force.
Section 1:
Replication of Brown et al.
Basic Single Equation Model.
Ta begin our analysis. we examine the 1mpact of the MW on TLFP in
the basic model.
The basic equation is as follows:
TlfP • dO + d]MW + dZPRIMEUR + d SY
AfP
POP
3
+ d
+ dSEfTPP + d
+
4
6
d7T + deTSO + d90Z + d1003 + d1104 + t
Sfnce we prov;ded considerable deta11s in Part 1. we will present
concisely the results of the basic equation.
We will do so in
different functional forms:
OLS l1near. GlS l1near. OLS logarithmic
and GlS logarithmic.
Similarly. bec au se the different specifications presented in
Part 1 did not brinq any major changes in the regression coefficients,
we will not undertake the estimation of the d1fferent specifications
and will concentrate on the basic equation as we d1d in Part II.
The tables below present the empir1cal results and are followed
byan explanatory comment.
94
Table 35:
Est1mated impact of an ;ncrease in the MW on the TLFP:
Basic OLS l1near.
hLFP = 0.13 - 0.161451MH - 0.554PRIMEUR - 0.4455Y + 0.151AFP -
(3.42)
(3.13)
(4.31)
(2.35)
0.568POP + 0.111EF1PP - 0.000651 + 0.0000226150 + 0.061102 +
(.89)
(.38)
(1.15)
(5.60)
(15.19)
0.12103 + 0. 0219 04
(21.25)
(5.18)
R2 = 0.961
F(ll: 92)
210.53
'R2 = 0.951
OH = 0.842
Table 36:
Estimated impact of an increase in the MW on the TLFP
Basic GLS L1near
LFP = 0.123 - 0.162424MW - 0.4325Y· 0.636PRIMEUR + 0.918AFP -
(2.39)
(2.50)
(3.01)
(2.04)
0.814POP - 0.325EF1PP - 0.0000161 + 0.0000198150 + 0.6102 +
(0.&4)
(1.58)
(0.013)
(2.31)
(19.34)
0.12403 + 0.021904
(31.32)
(6.18)
R2 = 0.914
F(ll: 91) = 301.103
'R2 = 0.911
OH=1.991
95
Table 31:
Est1mated impact of an increase in the MW on teenage labor
force participation - Basic OLS 10gar1thmic.
lnTLFP = -1.11 - 0.10041nHW - 0.191nPOP - 0.04631nPRIHEUR - 0.995Y +
(3.10)
(1.21)
(3.17)
(4.10)
1.16AFP - 0.328EFTPP - 0.000343T + 0.000393T50 + 0.12902 +
(1 .43)
(0.46)
(0.119)
(3.14)
(13.80)
0.24403 + 0. 04804
(23.54)
(4.99)
R2 = 0.941
F(ll: 91) = 151.46
ji2
0.941
DW = 0.943
Table 38:
Estimated impact of an increase in the MW on the tee nage
labor participation;
Basic model:
GLS logarithmic.
lnTLFP = -1.16 - 0.1215881nHW - 0.3511nPOP - 0.06311nPRIHEUR -
(2.819)
(1.20)
(3.36)
1.0025Y + 1.61AFP - 1.46EFTPP + 0.00266T + 0.000015150 +0. 12602+
(2.53)
(1.35)
(2.90)
(0.83)
(1.11)
(16.13)
0.24603 + 0. 045 04
(25.003)
(5.20)
R2 = 0.962
F(11: 91) = 202.14
ji2 = 0.951
DW = 1.99
RHO = 0.58
t(RHD) = 1.34
The results displayed in Table 34 through 31 above portray the
effect of an increase in the MW on labor force participation of youths.
In sharp contrast to the mixed reaction observed in the teenage
unemployment equation, the reqression coefficients unambiguously
demonstrate that an increase in the MW leads to a decline in the labor
9&
force participation of youths; that 1s a w1thdrawal fram the labor
force.
Moreover, the observed w1thdrawal 15 statistically sign1ficant
cons1stently (t-value : 3.42).
For instance. the result 1n Table 37 (GlS logarithmic) shows that
a 10 percent increase 1n the MW is expected to y1eld a 1.27 percent
withdrawal from the labor force by youths.
Al' of the above mentioned results concern only the basic
equation.
la assess the effect of a joint increase in FlS. WI and MW,
we turn to the extended model.
Section 2:
Extension of the Basic Teenage Laber Force Participation
Equation:
Inclusion of Female Laber Supply
and Wage-Inflation
We repeat the schemes of analysis adapted earlier by presentlnç
the results in Tables 38 through 41.
The most salient features of
those tables are summarized in an explanatory commentary.
Table 39:
Extended basic model:
estimated impact of an increase in
the MW, FlS and WI on the teenage labor force participation
rate.
OLS 1inear.
LFp. 2.&0 - 0.151458MW - 2.02PRIMEUR - 0.358SY + O.l&&AFP - 3.lJPOP
(3.&0)
(7.18)
(4.01)
(0.57)
(4.57)
0.0&8EFTPP + 0.000395T + 0.00001173TSO + 0.04&102 +0. 10303+
(0.2&)
(0.742)
(2.99)
(10.25)
(20.14)
0.0053404
1.8733lFLS + 0.09&9WI
(1.11 )
( 5. 85)
(0.763)
R2 = 0.973
F(13; 89) = 249.28
\\i2 = 0.969
OW = 0.98
97
Table 40:
Extended basic mode1:
estimated impact of an increase in
the MW. FL5 and WI on the teenage Tabor force participation
rate:
6LS I1near.
LFP " 2.25 - 0.0937268MW - 1.56PRIMEUR + 0.248AFP - 0.3425Y -
(1.51)
(5.74)
(0.562)
(2.26)
0.454EFTPP - 1.57POP - 0.00116T + 0.000022T50 + 0.05102 +
(2.46)
(1.36)
(0.98)
(2.83)
(14.11)
0. 11203 + 0.011304 - 1. 63886FL5 + 0.0390604WI
(24.88)
(2.80)
(5.04)
(0.483)
R2 = 0.980
F(13; 88) = 307.77
jl2 = 0.978
OW = 2.047
RHO = 0.61
t( RHO) = 7. 87
Table 41:
Estimated impact of an increase in the MW, FLS and WI on
the teenage Tabor force participation rate:
OLS
logarithmic.
lnTlFP = -3.85 - 0.1163941nMW - 0.1711nPRIMEUR - 0.7941nPOP -
(4.79)
(9.14)
(6.84)
0.5955Y - 0.946EFTPP + 0.00261 + 0.000014T50 + 0.094 702 +
(3.02)
(1.69)
(2.15)
(1.73)
(10.48)
0.20103 + 0.009704 - 3.896681nFl5 +0.18715IWI
(19.74)
(1.014)
(7.51)
(0.68)
R2 = 0.968
F(12; 90) = 230.99
jl2 = 0.964
DW = 1.25
98
Tabl. 42:
Est1mated effect of an 1ncrease 1n FL5, WI and HW on
teenage labor force participation:
GL5 10garlthmlc
'nTLFP· -3.63 - 0.019774011nMW - 0.1621nPRIMEUR - 0.7721nPOP -
(2.75)
(6.78)
(3.47)
0.6385Y - 0.651AFP - 1.64EFTPP + 0.002031 + 0.0000183T50 + 0. 09802 +
(2.25 )
(0.686)
(3.42)
(0.921)
(1 .24)
(10.64)
0. 2080 3 + 0. 0129 04 - 3.605601nFL5 + 0.1J4304WI
(18.51)
(1.30)
(5.07)
(0.813)
R2 • 0.970
F(13: 88) = 194.30
2 • 0.905
ow = 1.98
HO = 0.45
t(RHO) = 4.61
Three major observations emerge From the analysis of the extended
mode1 of the labor force participation rate of youths.
1- The HW index variable continues to demonstrate a sign1ficantly
negative impact on teenage labor force participation.
2 - The female labor supply variable proved to have a negative
impact on teenage labor force participation.
Furthermore. the impact
;5 s1gnif1cant and of greater magnitude than the impact derived From
an increase 1n the ~.
3 - The wage-inflation variable positively affects youth labor
force participation; but 1ts effect is ins;gnificant.
For instance, in the OLS logarithmic estimation (Table 40), an
1ncrement in both the HW and FLS produces a s1gnif1cant withdrawal
from the labor force participation of youths.
A 10 percent increase
in the MW leads to a 1.16 percent withdrawal from the labor force,
whereas a 10 percent 1ncrease in FLS leads to about a 38 percent
reduction in youth labor force participation.
99
All of the abo~e results stem from the extended basic single
eQuation .odel. where FLS and WI are ta ken as predetermined.
To 1nclude FLS and WI as endogeneous for reasons cited earlier.
we take up the simultaneous equations analysis.
Section 3:
A Simultaneous Equations Analysis of the Extended
Teenage Labor Force Participation Model.
Two Stage Least
Squares and Instrumental Variables Methods
In this section are reported the results of the joint estimation
of the TLFP eQuation.
Following the results displayed in the tables
below, we provide a concise explanatory comment.
Table 43:
Est1mated impact of an increase in the MW. FLS and WI on
teenage labor force participation:
2 SLS method.
LFP-2.17 - 0.110469MW - 1 .12PRIMEUR - 0.488SY + 0.515AFP -
(2.65)
(6.01)
(5.41)
(1.91 )
0.198EFTPP - 1.44POP - 0.00067T + 0.0000162TSO + 0. 05602 +
(0.748)
(2.54)
(1.31 )
(4.55)
(14.29)
0. 121 03 + 0. 0209 04 - 1. 53688F LS + 1.19432W!
(31.68)
(5.55)
(5.70)
(2.94)
R2 _ 0.974
F(13: 89) = 265.39
2 = 0.971
OW=1.17
100
Table 44:
Est1mated impact of a joint increase in the MW. FLS and WI
on teenage labor force participation:
2 SLS GLS method.
~LFP • 1. 94 - 0.110911MW - 0.995PRIMEUR - 0.4925Y + 0.615AFP -
(1.86)
(4.36)
(3.68)
(1.56)
0.408EFTPP - 0.91POP - 0.0009011 + 0.000019150 + 0.05102 +
(1.93)
(1 .03)
(0.96)
(3.11)
(15.25)
0.12303 + 0.02104 - 1.31255FL5 + 0.116201Wl
(32.03)
(6.05)
(3.10)
(1.16)
R2 = 0.916
F(13, 88) = 254.61
~2 = 0.913
DW = 1.86
Table 45:
Estimated impact of a joint increase in the MW, FLS and WI
on teenage labor force participation:
2 SLS-logarithmic.
ln1LFP = -2.50 = 0.028191nMW - 0.1021nPRIMEUR - 0.5011nPOP - 1.1225Y +
(0.954)
(6.433)
(3.46)
(5.31)
0.51AFP - 1.49EF1PP - 0.0009461 + 0.000028150 + 0.1102 + 0.24603 +
(0.85)
(2.35)
(0.69)
(3.09)
(11.59
(21.99
0.04804 - 4. 1362fLS + 1. 93245WI
(5.68)
(5.93)
(2.005)
R2 = 0.965
F(13: 89) ·191.38
~2 = 0.960
DW = 1.21
An analysis of the figures in Tables 43 through 45 leads to three
major observations.
l - The estimated impact of an increase in the female labor
supply is consistently statistically significant in the joint
estimation, when the eQuat10ns are either in l1near or logarithmic
fonn.
101
Z - The MW index on the other hand, conveys m1xed infonmation.
The est1mated impact on an 1ncrement in the MW proved to be
sign1ffcant in the 2 SlS wh en the equat10ns are in the lfnear form.
but becomes fnsignfffcant in the logarithmic form.
3 - The wage-fnflation variable affects positively the teenage
labor force participation; that îs. an increment 1n wage is likely to
entice teenagers to keep searching for jobs, rather than withdrawing
From the labor force.
As a final test of how robust are the 2 SlS coefficients, we
estfmated the joint model by applying the instrumental variables
method.
The results derived From the instrumental variables method
are dfsplayed in Appendix G.
These results do not dfffer
substantially from those obtained in the 2 SlS estimation.
We summarize all of the results of the teenage labor force
participation in Table 46.
the regression coefficients have been
converted into elasticities.
Table 46:
Estimated effect of a 10 percent increase in MW. FlS and WI
on teenage labor force participation (in percent).
~peciflcations and/or
Effect of
Effect of
Effect 0
Functional Forms and
~
fLS
WI
Methods of Estimation
l - Baslc - OLS linear
~!.o~~
NI'
NI'
3.42
2 - Basic - GlS linear
-0.994
NI
NI
(2.39\\
3 - Basic: OLS logarithmic
-1.004
NI
NI
(3.10\\
102
4 - Basle GLS logar1thm1e
~:.2;~
NI
NI
2. Bl
5 - Extended Basic - OLS
-0.92
-~1.9~\\
;~.02~\\
l1near
13.60\\
5.B5
0.763
6 - Extended Basic - GlS
-0.57
-27.96
.0.011
1inear
Il.51\\
(5.041
(0.4631
7 - Extended Basic: OLS
-1.16
-~6.9~\\
;~.o~~
loaarithmic
14.79\\
7.51
0.66
6 - Extended Basic: GLS
-0.66
-~5.6l1
loaarithmic
f2.75\\
;~.5;\\
5.67
0.61
9 - 2 SLS Extended Basic
t6~\\
-~6.2~1
;~.3~\\
2.65
5.70
2.94
0 - 2 SLS - GLS
-0.67
-22.32
.0.23
Estended Bas;c
Il.661
13.701
(1. 761
1 - 2 SlS - logarithmic
-0.26
-41.36
.0.57
Extended Basic
(0.95)
(5.93)
12.0051
2 - 2SLS Extended Basic
-0.55
-31.96
.0.05
lOQarithmic
Cl .36)
(4.07\\
Cl.66)
3 - Instrumental Variables
-0.627
-46.71
>0.0692
Method: OLS linear
12.90)
(5.651
10.691
4 - Instrumental Variables
-0.166
-12.67
0.0172
Mothod GLS 1100ar
1] .62 \\
(2.931
10.731
*Mon 1ncluded.
Table 4b presents the est1mated effect of a la percent increase
in the minimum wage index, female labor supply and wage-inflation on
the teenage labor force participation, that 15 those teenagers who are
employed or look1ng for work.
A close examination of the elasticity coefficients reveals sorne
outcomes of 1nterests.
First, 1n lines 1 through 4 are the estimated effect of an
increase 1n the MW on teenage labor force. when the estlmation 1s
carried out on the basic tee nage labor force model.
103
It cao be seen that a 10 percent 1ncrease in the HW produced
about a one percent dec11ne in the teenage labor force.
These effects
proved to be stat1st1cally sign1f1cant.
Accordingly. one m1ght state that increases in the MW cause a
significant withdrawal of youths from the labor force.
Furthenmore,
in line 1. a 10 percent increase in the HW gives fise to a 1.02
percent w1thdrawal from the labor force by youths.
This coefficient
of elasticity 1s ident1cal to the one obtained when we estimated a 10
percent increase in the MW on teenage employment (Table 11, line 1;
page 60).
Consequently, one might expect that in the basic model, the
unemployment effect of a 10 percent increase in the MW would be zero,
since the teenage labor force withdrawals and the teenage employment
reductions are equally proportional.
Indeed, the teenage unemployment
rate, though not exactly equal to zero. is practically close to zero.
Such a gap may stem from the fa ct that certain teenagers drop-out of
the labor force out of d1scouragement.
Those d1scouraged workers are
no longer counted as unemployed. and are at the core of the imprecise
nature of unemployment as an accurate descript1ve measure of labor
market conditions.
The second observation one can make from Table 45 emerges from an
examination of l1nes 5 through 8, where the extended teenage labor
force participation model is analyzed.
The HW index continues to show a significant withdrawal from the
labor force; a 10 percent increase in the HW produces about 0.57
l~
percent w1thdrawal From the labor force by youths; the impact is
reduced compared w1th the basic model.
Female labor supply demonstrates a surpris;ngly strong impact on
the youth labor force~ we expected the negative sign, not the
magnitude.
A 10 percent 1ncrement in FlS g;ves rise to about a 28
percent withdrawal From the labor force.
ln addition, the FLS impact
proved to be statistically significant.
In sharp contrast to the FlS impact. the wage-inflation variable
exhibits a positive effect on the teenage labor force. although this
effect proved to be insignificant.
The third observation that cornes out of Table 45 is concerned
w1th the analysis of the elasticity coefficients of the policy
variables when the simultaneous equations estimation is applied. as
shown in lines 9 through 14.
The MW index maintains its adverse
effect on teenage labor force; but its statist1cal significance is
questionable.
Indeed in line 11, one realizes that when the
10gar1thmic fonm 1s used, the HW index effect 1s no longer sign1ficant
statistically.
Accordingly. one might state that the ~ demonstrates
some ambigu1ty in the s1gnificance of 1ts negative impact on teenage
labor force.
More 1nterest1ng, the above ment10ned ambigu1ty seems to
lend support to the amb1guous impact of the MW on teenage employment
encountered in section 3 of part 1.
Another salient point we note is that in the 2 SLS, the
wage-inflatton tmpact proved to be significant, although such impact
1s still small in magnitude.
This implies that an 1ncrease in the
wage rate mot1vates teenagers to search for jobs.
105
Perhaps the most str1k1ng result 15 the consistently strong
negat1ve impact of an 1ncrease in FLS on the teenage 1abor force.
A
10 percent 1ncrease 1n the FLS in the joint estimation produces about
a 22 percent withdrawal of youths From the labor force.
Although we
found such an impact surprisingly strong. the direction of the effect
was expected.
Indeed, most previous studies have found appreciable labor-force
w1thdrawal in response to increases 1n the minimum wage index.
Dur
findings therefore 1end support to past studies.
However, our study sheds new 11ght in that it includes the female
labor supply and wage-inflation variables.
By incorporating the FlS and WI, we d1sclosed that the withdrawal
generated by an increase in the HW might not be significant
cons1stently. whereas the FLS demonstrates a stronger and consistent
negative effect on· the youth labor force.
ln addition. 1t should be noted that 1n the light of the
different functional fonms experiments we conducted, along with the
various estimation methods, we found our results quite robust.
106
CHAPTER IV:
SUMMARY AND POLICY IMPLICATIONS
The main objective of th;s research is to shed new 11ght on the
controversy surround1ng the impact of 1ncreases in the federal minimum
wage on the labor marKet.
Special attention is directed at the
teenage labor force status, as concern about youth unemployment has
grown and because the minimum wage is often blamed for 5uch high youth
unemployment.
80th past studies (with the Brown et al. work as
reference) and the findings of thls present research are presented.
We relate our results to prevlous ones.
To be comparable with other
studies, quarterly data were used.
Most previous studies confined themselves to examining the effect
of changes in the m1n1mum wage on teenage employment or unemployment
rates.
We analyzed the effect of increases in the mandatory wage not
only on employment. but also on the unemployment rate and on the labor
force participation rate, following the Brown et al. framework.
However, the most important feature of our study is that for the
first time the analysis is extended to incorporate new explanatory
variables. namely the female labor supply and wage-inflation.
Motivat1ng this extension 1s the observed sharp increase in the
labor force participation rate of women, which constitutes the most
fundamental change in the labor market since World War Il.
Such changes in the composition of the labor force must be part
of any comprehensive analysis of the unemployment level of any segment
107
of the total labor force.
L1kewise. the cont1nuous increases in the
average hourly earnings must also be part of an evaluation of the
impact of the mandatory wage on the labor market.
Nevertheless, no previous study has explicitly and systematically
cons1dered the impact of increases in the female labor supply and
wage-inflation in the analysis of the youth unemployment issue.
In
that sense alone, our contribution is in suggesting a new and broader
approach ta better understand the factors that lie behind youth
unemployment and/or withdrawal From the labor force.
First we introduced bath fernale labor suppl y and wage-inflat;on
as exogeneous variables.
But it is well-known that bath variables are determined themselves by
relevant socio-economic factors.
Consequently. we incorporate at a second stage both variables in the
model as end0geneous. and a simultaneous eQuat10ns procedure is then
utilized to assess the impact of both variables on teenage
employment.
This estimation technique 1s appl1ed to each definition
of the labor force status.
The results show great sensitivity to the ordinary and
generalized least squares techniques of estimation.
But because of
the presence of serial correlation. we put more weight on the
generalized least squares results.
Most previous studies reported only the ordinary least squares
estimates and found that a 10 percent 1ncrease in the minimum wage
leads to a one to three percent reduction in teenage employment.
We
108
round that a 10 percent 1ncrease in the minimum wage y1elds a zero to
one percent re~uct1on in teenage employment in the basic equat1on.
This supports the Brown et al. ftnd1ngs.
The lower numbers are
obta1ned from the genera11zed least squares estimates and linear
relattonsh1p between teenage employment and the minimum wage.
The
higher estimates are derived from the double-logarithmic equations.
But this negative impact seems significant only in sorne
specification and/or functional fonms.
However, once the model is extended to include the female labor supply
and wage-inflation variables, we note considerable changes in the
outcome.
The negative impact of the minimum wage all but vanishes and clearly
is no longer statistically significant.
In contrast, the female labor
supply does exhibit a strong negative and significant effect on
teenage employment.
The estimates imply that a la percent 1ncrease in the female
labor supply leads to about a 28 percent decline in youth employment.
This female labor supply effect proved to be robust w1th regard to the
alternative funct10nal forms and specifications uti1ized.
On the other hand. wage-1nflat10n shows a positive but insignificant
effect on youth employment.
As discussed above, we also explored the behavior of the policy
variables using s1multaneous equations procedures.
Once aga in,
important changes appear in the outcome of the basic model.
The
minimum wage has a small and ins1gnificant ~mpact on teenage
employment.
Also, wage-inflation shows a positive lnslgnificant
effect.
109
Only the 1ncreases in the female labor suppl y demonstrate a
consistent and sign1ficant negative impact on teenage employment.
A
10 percent increase in the MW and the FLS produced 0.34 percent and
26.83 percent reduction 1n youth employment respectively.
In
contrast. a 10 percent increase 1n the wage inflation gives rise to
about a 0.06 percent ;ncrease in teenage employment.
Similar results are obta;ned when the teenage labor force
participation rate is used as dependent variable.
Wage-inflat1on has a negligible and positive effect on teenage labor
force participation.
Both the female labor supply variable and the minimum wage index are
inversely related to the participation of youths in the labor force.
Both exhibit a statistically significant inverse relationship to the
labor force participation rate, though of different magnitude.
A 10 percent increase in the minimum wage index led to a
reductton From O.BD percent to 1.61 percent decline in the
participation rate of youths; whereas a la percent increase in the
female labor suppl y reduces teenage labor force participation rate by
about 30 percent.
Like most past studies, we found sufficient w1thdrawal From the
labor force of youths as a result of an increase in the minimum wage.
Hevertheless our find1ngs show that the withdrawal resulting From
an increase in the female labor suppl y is of greater magnitude.
110
Prev10us stud1es accepted the surpr1singly large w1thdrawal of
teenagers frQa t~e labor force w1thout any explanat10n other than the
negative effect of the minimum wage.
But when one notes that the
influx of female workers into the labor force has grown From one third
before World War II to 1t5 current level of 53 percent, one might
state that the increase in the female labor suppl y is an underlying
factor to take into consideration in assessing the detenminants of the
withdrawal of teenagers from the labor force.
This hypothesis does not necessarily indicate a causal
relationship between teenage labor force participation rate and female
labor supply.
Rather, it is consistent with the strong negative
correlation between these two groups of workers. which can explain
"the extent to which women and youths are substitutes in production~
as Hamenmesh and Grant have observed.
The results are somewhat more difficult to interpret when one
uses the teenage unemployment rate as a dependent variable.
The
estimated effect on the teenage unemployment rate of increases in the
policy variables are either very small or wrong-signed.
Furthenmore, those estimates exhibit great imprecis10n as shown
in the very high standard errors in the basic equation.
The extended
model provides no substantive changes relative to the basic model.
The female labor suppl y variable proved to have the only
significant result but its point estimate implies a negative rather
than 1ts expected positive effect.
111
The wage-1nflat1on variable has almost no d1scern1ble impact
under all the d1fferent exper1ments.
Despite these shortcomings, as a
whole. one might cautiously state that a la percent 1ncrease in the
minimum wage would increase teenage unemployment rate by about one
tenth of one percentage point.
In stark contrast to the small minimum wage impact, a la percent
change in the female labor suppl y yields a nine to about thirty
percentage point change in the teenage unemployment rate, but in the
wrong direction.
Taken together. these results lend credence to the accepted
notion that the employment rate and/or the labor force participation
rate, are better descriptive measures of the conditions of the labor
market than is the unemployment rate.
Policy Implications
The findings reported 1n this research appear to have important
policy implications.
Indeed, one of the most freQuently suggested
proposals to deal with the high tee nage unemployment rate consists of
a "subminimum- or -dual min1mum- wage for teenagers.
The results of th1s study suggest that such a policy proposal, if
applied in isolation, might not prov1de the expected remedy for the
overall teenage unemployment.
Host important, Dur findlngs might put lnto doubt the adeQuacy of
the submlnlmum wage proposal as a viable po11cy solution for high
teenage unemplo~nt.
112
On the one hand. our results have demonstrated that an 1ncrease
in the female labor supply variable has cons1stently shawn a strong
and stat1sticallY s1gn1f1cant negat1ve impact on the teenage
employment and on the teenage labor force participation.
On the other hand. the minimum wage has shown sorne amb1guity in
its adverse effect on bath teenage employment and on the teenage labor
force, when the model is esttmated using the simultaneous equat;ons
procedure.
As for the wage-inflation variable. 1t has had a
consistent positive impact but the impact has been 50 small that it is
appropriate to state that it has proved not to have any meaningful
effect on youth employment and labor force participation.
Therefore, given the incertitude surrounding the level of
significance of the impact of increases in the minimum wage on youth
employment and labor force, and given the cons1stency of the female
labor supply's sign1f1cant negative impact on youth employment and on
labor force participation, one should cons1der the "youth
differential" minimum wage proposal with caution.
Before we analyze the different aspects of the "dual minimum-
wage proposal. we will explain what -subminimum" wage really means.
Conceptually. a "youth differential" min;mum wage. its advocates
believe, would reduce the wage employers could pay to youths.
This
will have the effect of making employment for the unskilled teenagers
more feasible to employers who could pay youths a wage below the
national minimum wage and still comply with the Fair labor Standards
Act of 1938 wh1ch established the national minimum wage.
113
ConseQuently, continue the proponents. the dual minimum would help
reduce the persistently h1gh teenage unemployment rate.
ln its simplest fonm, the subminimum wage would al10w employers
to pay youths about 1S percent or es percent of the basic minimum wage.
In stark contrast, crit;cs contend that 5uch a dual minimum wage
would be at the expense of older high paid workers.
The prospect of
the possible substitution between teenagers and adults ;5 at the heart
of the controversy surrounding the dual minimum proposal.
"issing in th\\s debate, however, is the factor capturing the mast
fundamental change that has occurred over time in the American labor
mark.et:
the increase in the female labor supply.
ln fact, the
tremendous influx of women into the labor force cannot be ignored if
one is to conduct a comprehensive analysis of the employment
conditions of one segment of the labor force.
For the first time. we
introduced the female labor suppl y explicitly into the analysis.
Our
results have clearly disclosed that an increase in the female labor
supply variable does produce a greater decline in youth emp10yment or
w1thdrawal From the 1abor force than the "W.
This lends credence to
Hamenmesh and Grantls (19Bl) findings that teenagers and women are
close substitutes in low-skilled production activities.
Under these circumstances, a mere lowering of the minimum wage
for one special group--teenagers--could not reduce substantially their
unemployment rate.
Opponents often contend that the dual minimum would not generate
add1t1onal jobs for youths.
The reasons they failed to provlde stem
From the chang1ng structure of the composition of the Tabor force.
114
Certa1nly a lower minimum wage would 1nduce fa~t-food employers
to h1re a greater number of youths.
But for the economy as a whole.
not all the teenagers will be absorbed if the substitution between
youths and women takes place.
For example. the fast food chain
Stake-n-Shake responded to the 1978 jump in the minimum wage by
replacing teenagers with fewer older women workers who tend to be more
productive, and easier to train and sa create less turnover.
Several
factors have contributed to the increase in female labor supply.
One
of the most important is the lowering of the reservation wage of
women.
Indeed, prier to World War Il, women had a high reservation
wage, ma1nly because they valued then non-market tlme to perform
traditional household work more than market time activities.
Other
factors include the changing nature of jobs (fewer heavy-duty
manufacturing jobs and more service jobs). the increased opportunity
of education for women. the development of child care services, and
the changes in attitudes (acceptance of more single women, the
emergence of the women's movement, the fight aga;nst sex
discrimination in the work place).
Al1 of the above factors along
w1th the insufficient incorne of husband. have decreased the value of
non-market time for women and consequently lowered their reservation
wage.
This decline in the women's reservation wage relative to the
minimum wage whlch did change upward. has prompted women to enter the
labor force, thereby increasing the total labor pool of the economy.
Therefore, if the subminimum wage ;s to reduce youth unemployment
as expected, 1t must be set 50 that it 15 below the women's
......-
115
reservation wage.
If th1s 1s done. it w1l1 encourage women to
w1thdraw from the 1abor force.
Teenagers may then f111 in the
open1ngs created by these w1thdrawals.
But at that level. the
subm1nimum wage will be 50 low that only migrant workers without
reservation wage. may be willing ta accept to work.
It should be noted at this point that the withdrawal of women
From the labor force should not be a deliberate policy option.
Still,
if the dual minimum wage is not below the women's reservation wage.
women will continue ta supply a great percentage of the labor force.
Substitution between youths and women will occur and the "dual
mi nimum" wage wi 11 become an 1mpotent po li cy opt i on.
Furthenmore. Hamermesh and Grant showed that women proved to be
preferred subst1tutes for teenagers by many employers.
Such favorable
treatment of wornen turned out to prevail even in sorne jobs where
existlng wages different1als are in force under section 14 of the Fair
Labor Standards Act.
This is the case for full-time students,
learners not 1n retail or service industries, and handicapped workers
in sheltered workshops. whom employers are authorized to pay less than
the minimum wage.
Should th1s preferential treatment be motivated by
the practice of role prejudice of employers, then antidiscrimination
policy measures might be necessary.
ln addition, teenagers do not constitute a homogeneous group.
Many studies have demonstrated that minor;ty teenagers' unemployment
;s at least tw1ce as h1gh as their white counterparts.
Hence, a
subm1n1mum wage alone cannot solve the high overall teenage
116
unemployment if 1t 1s 1ndeed the outcome of labor market
discrimination.
As a whole. we find that 1ncreases in youth employment do not
translate into a one-for-one reduct10n 1n youth unemployment.
This
confinms perv10us studies.
Beyond that point, our contribution
res1des 1n proposing a broader framework of analysis than the Dnes
encountered in past research.
We did sa by incorporating the female
labor suppl y into the analysis.
This inclusion ra;ses questions about the adeQuacy of the dual
minimum wage to curb the high teenage unemployment.
Moreover, lt
prov1des sorne information concern;ng the factors that underlie the
discrepancy between teenage employment increases and reductions in
teenage unemployment.
At the outset, we favor a policy of an overall sustained economic
growth that will enable the economy to absorb the increased labor
force, rather than a policy option whose ultimate effect is to
confront one segment of the labor force with another.
In the tight
labor market situation stemming From the overall economic expansion,
the wage rate w11l be based on jobs, not on the characteristics of the
people who hold these jobs.
117
End Notes of Chapter III
11 am indebted to Professor C. Brown for making his data set
dvailable to us.
2Citibase 15 made dvailable to us by the Social Science
Research Facilities (SSRF), Department of Economies of the University
of Wisconsin-Milwaukee.
30ata on years of school completed were obtained From the
-Labar Force Statistics· derived From the CPS:
A Data Book, Vol. l,
p. 7751.
One should note that the Question on educational attainment
app11es only to progress 1n -regular" schaols.
5uth schaols include
graded public, private and parochial elementary and high schaols.
colleges. un;versities and professional schaols. whether day school or
night schools.
The median years of school is def1ned as the value which divides
the distributlon into two equal groups; one having completed more
schooling and one having completed less schoollng than the median.
These medians are expressed in terms of a continuous series of numbers
representlng years of school completed.
For example, a median of 13.0
means completion of the first year of college.
41 am grateful to Mr. Howard Hayghe of the U.S. Department of
Labor. Bureau of Labor Statistics (BLS) for ma~ing the series
available to our use.
5The median incorne is the amount whlch d1vldes the distribution
of incorne into two equal groups:
one havlng incarne above the median
and the other having below the median.
Data on husbandls income were obtained from the U.S. Bureau of
Census Publlcations titled -Current Population Reports. series P-Bo,
Income of Households. Famllles and Persons."
6The median incomes were published annually.
To obtain
quarterly data w1thout 10s1ng the tlme trend. a spline functions
technique was used.
professor J. W. E11iott provided us with the
computer assistance availab1e at the Massachusetts lnstitute of
Technology.
7Children are defined as own children and related children .
• Qwn- children in a faml1y are sons and daughters. including
stepchildren and adopted children of the family head.
·Related" children in a family include own chi1dren and al1 other
chl1dren in the household who are related to the family head by b10od.
marri age or adoption.
lhe data were obtained from natality statistics analysls in the
United States 1917-197B monthly vital statlstics report.
CPR series
P-20, No. 336. April 1919.
As 1n MOst studies, the proxy of total
fert11ity rate was used.
118
Total fert111ty = sum of age specifie ferti11ty rates for single
women. 14-49 years of age.
Th1s represents an estimate of the number
of ch11dren barn per woman over the child bear1ng period.
80LS:
ord1nary least squares.
9GLS:
general1zed least squares correction of serial
correlation of the error by the Cochrane-Orcutt Iterative Technique.
10]n logarithmic equat;ons. TEMP. MW, PRIMEUR. POP and PCWEL
are entered logarithmically; other variables ]inearly .
•
11Using arder-condition and Kmenta Rank Condition, we find out
that the system was overidentified.
This prompted us to use the
two-stage least square instead of the indirect least square method.
12Suc h ambiguity is supported by the multiple correlation
coefficients presented in Appendix H.
119
APPENDICES
120
ApDendlx A:
Basic Teenage Employment Equat1on.
Regression Results.
OLS llnear; Different Specification.
- BASIC - TSO
EMP = 1.09& - 0.133B5BMW - 1.429PRIMEUR - 0.&97SY + 0.5BB2EFTPP -
(2.5&)
(9.101)
(&.59)
(1.92)
3.&02POP + 0.023AFP + 0.00172T + 0.03502 + 0.9&03 + 0.014204
(10.3)
(0.073)
(&.B7)
(8.22)
(20.5)
(3.1&)
R2 = 0.940
F(10; 93) = 14&.138
R2 = 0.933
DW = 0.84
- BASIC - SY
EMP = 0.45 - 0.099&B59MW - 1.2&7PRIMEUR - 0.&27AFP + 0.3&IEF1PP -
(1.909)
(7.91)
(1.88)
(1.12)
0.2&lPOP - 0.001721 + 0.00002&7S0 + 0.03702 + 0.100503 + 0.01104
(0.38)
(2.99)
(&.3&)
(B.72)
(20.80)
(3.7&)
R2 = 0.93B
F(10; 93) = 142.79
R2 = 0.932
DW = 0.7015
121
- BASIC - AFP
EHp· 0.87 - 0.11316HW - 1.36PRIHEUR - 0.55SY + 0.B7EFIPP - 1.45POP-
(2.54)
(9.85)
(5.61)
(0.304)
(2.82)
0.000741 + 0.0000171S0 + 0.3602 + 0.9903 + 0.15404
(1.39)
(4.96)
(9.56)
(23.6)
(3.85)
2 • 0.952
F(lO; 93) • 187.36
2 • 0.947
DW· 0.786
- BASIC
EFIPP
EHP· 0.79 - 0.1411HW - 1.323PRIHEUR - 0.55SY + 0.61AFP - 0.783POP -
(3.10)
(9.61)
(5.85)
(2.05)
(1.34)
0.000941 + 0.00002091S0 + 0.03702 + 0.10103 + 0.01504
(1.80)
(5.89)
(9.93)
(25.5)
(4.03)
2 • 0.954
F(10; 93) • 196.07
2.0.949
DW·0.786
122
AA51C - pop
TEMP = 0.11 - 0.1430bMW - 1.24PRIMEUR - 0.5215Y + 0.821AFP +
(3.115)
(9.b8)
(5.51)
(3.23)
0.0033EF1PP - 0.001551 + 0.000025150 + 0.03802 + 0.10203 +
(0.012)
(5.bO)
(12.92)
(10.29)
(25.34)
0.01b904
(4.33)
R2 = 0.953
F(10; 93) = 192.1b
fi2 = 0.948
OW = O.lbb
•
BA51C + PCWEL
EMP = 0.89 - 0.1544MW - l.b99PRIMEUR - 0.5b15Y + 0.5bAFP - 0.34EF1PP
(3.89)
(12.33)
(b.18)
(2.15)
(1.33)
1.158POP - 0.0004011 + 0.000014150 + 0.03302 + 0.09803 +
(3.23)
(0.8b)
(4.21)
(10.23)
(21.00)
0.011104 + 0.281PCWEL
(3.3b)
(5.02)
R2 = 0.9bb
F(12; 91) = 218.3b
~2 = 0.9b2
OW = 1.041
123
BASle + EOP
~EMP = O.Bbl - 0.104MH - 1.3BPRIMEUR - 0.45SY + 0.45AFP + 0.190EF1PP -
(2.10)
(9.74)
(4.2b)
(1.45)
(O.bb)
0.b74POP - 0.000321 + 0.000015150 + 0.03b02 + 0.09003 +
(1.13)
(0.52)
(3.23)
(9.bl)
(23.2)
0.01504 - 0.22EDP
(4.b3)
(1.7B)
R2 = 0.95D
F(12; 91) = 105.B4
~2 = 0.950
DH = 0.84
124
Appendix 8:
Teenage Employment Equat1on; Regression Results. GlS
l;near.
D1fferent Specifications.
1 - BASIC - TSO
EHP = 0.924 - 0.0548HW - 1.086PRIHEUR - 0.54SY - 0.056AfP -
(0.815)
(5.29)
(2.84)
(0.113)
0.60EfTPP - 2.9POP + 0.0011T + 0.0402 +
0.10B03 +
0.0204
(3.52)
(4.5)
(13.8)
(29.1)
(29.14)
(6.19)
R2 = 0.911
f( 10; 91) = 389.62
R2 = 0.974
DW = 2.038
RHO = 0.801
t(RHO) = 13.53)
2 - BASIC -SV
TEHP = 0.42 - 0.37HW - 1.05PRIHEUR + 0.34AfP - 0.56EfTPP + 0.036POP -
(0.55)
(5.09)
(0.6B)
(3.20)
(0.021)
0.0019T + 0.000027TSO + 0.0402 + 0.10903 + 0.020804
(1.01)
(2.17)
(13.9)
(29.49)
(6.27)
R2 = 0.975
F(10; 92) = 365.93
R2 = 0.973
RHO = O.Bl
t(RHO) = 12.56
125
- BASIC - AFP
EHP • 0.79 - 0.049MW - 1.14PRIMEUR - 0.49SY - 0.57EF1PP - 1.21POP -
(0.7B)
(5.91)
(2.B3)
(3.21)
(1.01)
0.00071. 0.0000171S0 • 0.03902' 0.10703.0.01904
(0.49)
(1.B9)
(14.2)
(30.4B)
(6.11)
F(10; 92) = 371.49
• 0.973
OW = 1.92
t(RHO) = 10.52
- 8ASIC -EFTPP
[MP = 0.73 - 0.0712HW - 1.12PRIHEUR - 0.49SY • 0.50AFP - 0.59POP -
(1.12)
(5.83)
(3.03)
(1.13)
(0.501)
0.00116T • 0.0000204TSO • 0.03902 • 0.10403 • 0.01904
0.96)
(2.50)
(13.49)
(29.20)
(5.93)
2 = 0.972
F(10; 92) • 318.13
2 = 0.969
OW = 1.91
HO • 0.633
(RHO) • B.30
716
- BASIC - POP
EMP = 0.6B - 0.06061MW - 1.09PRIMEUR - 0.49SY + 0.55AFP - 0.56EFTPP -
(0.95)
(5.B9)
(2.91)
(1.39)
(3.24)
0.0016T + 0.000024TSO + 0.0402 + 0.10B03 + 0.0204
(2.7B)
(5.62)
(14.6B)
(31.29)
(6.41)
1 = 0.976
F(10; 91) = 372.9B
2 = 0.973
DW = 1.91
HO = 0.712
(RHO) = 10.31
- BASIC + PCWEL
EMP = 0.B2 - 0.090BMW - 1.36PRIMEUR - 0.54SY + 0.47AFP - 1 .2BPOP -
(1.59)
(7.2B)
(3.B7)
(1.1B)
(1.25)
0.5BEFTPP - 0.0007T - 0.000016TSO + 0.03702 + 0.10403 +
3.12)
(0.73)
(2.35)
(13.39)
(29.95)
0.01604 + 0.2225PCWEL
5.29)
(2.5B)
2 = 0.975
F(12; 90) = 275.13
2 = 0.971
HO = 0.57
(RHO) = 7.0B
127
1 - BASIC. [OP
TlMP ~ 0.79 - 0.0524MW - 1.16PRIMEUR - 0.42SY • 0.39AFP - 0.17POP -
(0.B2)
(5.99)
(2.39)
(0.B5)
(0.126)
0.51EFTPP - 0.000491 • 0.000016150 .0.03902.0.10603 •
(2.91)
(0.33)
(1.65)
(13.18)
(29.32)
0.01904 - 0.14EOP
(6.1B)
(1.69)
R2 ~ 0.911
F(12; 90)
299.42
R2 ~ 0.914
DW ~ 1.91
RHO ~ 0.11
(RHO) ~ 10.23
lZB
Appendix C:
Teenage Employment Equation; Regression Results.
OLS
logarithmic.
Different Specifications.
1 - BASIC - TSQ
lnTEMP· -2.95 - 0.105B1nMW - 0.1261nPRlMEUR - 0.B61nPOP - 1.53SY -
(3.0Z)
(9.69)
(10.47)
(6.06)
0.60AFP + 0.64EF1PP + 0.00451 + 0.090Z + 0. 2203 + 0. 0404
(0.15)
(0.88)
(1.84)
(8.96
(20.61)
(3.82)
R2 = 0.939
F(10; 93) = 143.49
il2 = 0.932
DW = 0.92
- - - - - - - - - - - - - - - - - - - - - - - - '
Z - BASIC - SY
lnTEMP = -2.13 - 0.01941nMW - O.llBlnPRIMEUR - 0.2531nPOP + O.SlAFP +
(2.15)
(8.38)
(1.41)
(0.61)
0.608EFTPP - 0.00Z11 + 0.000054150 + 0.09 402 + 0. 23 03 + 0.04404
(0.76)
(1.64)
(4.66)
(8.77)
(19.43)
(3.99)
R2 = 0.931
F(lO; 93) = 125.84
R2 = 0.923
DW=0.199
129
- BASIC
AFP
lnTEHP ~ -1.86 - 0.09b51nHW - 0.1201nPRIHEUR - 0.411nPOP - 1.30SY -
(3.14)
(9.99)
(3.21)
(5.36)
0.112EFTPP - 0.0006T + 0.000031TSO + 0.09302 + 0.2303 + 0.04304
(0.15)
(0.41)
(3.8b)
(9.91)
(22.5)
(4.4)
R2 ~ 0.941
FI10; 93) ~ 100.91
R2 • 0.941
OW • 0.903
- 8ASIC
EFTPP
lnTEHP ~ 1.11 - 0.10591nHW - 0.1111nPRIHEUR - 0.331nPOP - 0.3SY +
(3.1b)
(5.41)
(2.13)
(5.41)
0.T06AFP - 0.00091 + 0.000041S0 - 0.09402 + 0.2303 + 0.04404
(O.Bo)
(0.01)
(4.003)
(10.04)
(23.58)
(4.50)
2 ~ 0.941
F(lO; 93) ~ 220.11
R2 • 0.942
OW • 0.92
130
5 - BASIC - lnPOP:
ln1EKP' -0.88 - 0.112blnHW - 0.i071nPRIKEUR -
(3.40)
(9.30)
1.2bSY + 1 .74AFP - 0.29EF1PP - 0.003b1 + 0.0000b1S0 + 0.09802 +
(5.12)
(2.bl)
(0.409)
(4 .•S)
(11.4B)
(10.40)
0.2403 + 0.04704
(23.3b)
(4.B4)
R2 • 0.945
F(lO; 93)
jl2 = 0.939
OW = 0.843
b - 8ASIC + lnPCWEL:
lnlEKP = -1.81 - 0.15991nKW - 0.1491nPRIMEUR -
(5.95)
(13.98)
0.441nPOP - 1.28SY + 1.40AFP - 1 .70EF1PP = 0.0021 + 0.000041S0 +
(3.52)
( •.•• )
(2.13)
(2.83)
(1.87)
(5.24)
0.08702 + 0.2303 + 0.03404 + 0.10221nPCWEL
(11.57)
(28.24)
(4.38)
(7.45)
R2 = 0.9.7
F(12: 91) • 225.89
R2 = 0.9.3
OW • 1.33
- SASIC + EDP:
ln1EKP = -1.34 - 0.08251nHW - 0.131nPRIKEUR -
(2.50)
(10.09)
0.3031nPOP - 0.985Y + 0.093AFP + 0.305EF1PP + 0.00131 + 0.000022150 +
(1.98)
(3.72)
(0.112)
(0.43)
(0.79)
(1.85)
0.08902 + 0.2203 + 0.04204 - 0.78EDP
(9.59)
(21.22)
(4.33)
(2.62)
R2 = 0.951
F(12; 91) = 148.24
R2 = 0.944
DW = 0.97
131
Append1x D:
Teenage Employment Equation.
Regression Results.
GLS
Logar1thmic.
Different Specifications
1 - 8ASIC - TS9:
lnTEMPT, -2.94 - 0.09311nMW - 0.1221nPRIMEUR -
(1.98)
(6.11)
0.8151nPOP - 1.37SY - 0.094AFP - 1.84EFTPP + 0.0054T + 0.09592 •
(5.11)
(3.16)
(0.082)
(3.89)
(6.75)
(11.50)
0.2493 + 0.04594
(24.64)
(5.14)
R2 , 0.967
F(10; 92) " 272.90
ji2 • 0.964
DW" 2.07
RHO • 0.68
t(RHO) • 9.58
2 - 8ASIC -SV:
lnTEMP· -2.53 - 0.006851nMW - 0.1161nPRIMEUR -
(1.37)
(5.29)
0.4061nPOP + 0.15AFP - 1.82EFTPP - O.OOOOyuT + 0.00004TS9 +
(0.97)
(0.114)
(3.82)
(0.016)
(1.27)
0.09792 + 0.2493 + 0.04794
(11.60)
(23.27)
(5.07)
R2 • 0.967
F(10; 92) " 265.56
R2 • 0.963
OW " 2.053
RHO • 0.74
t(RHO) • 11.16
132
3 - BASIC - AFP
InTEMP· -2.1& - 0.08371nMW - 0.121nPRIMEUR - 0.Sl&1nPOP - 1.27SY -
-(1.87)
(&.34)
(1.91)
(3.0S)
1.80EFTPP + 0.0013T + 0.000043TSO + 0.09&02 + 0.2403 + 0.04S04
(3.77)
(0.3B)
(1.2&)
(12.&1)
(2S.43)
(S.34)
R2 • 0.9&7
F(10; 92) = 2&9.&7
jl2 = 0.9&4
DW = 2.02
RHO = 0.&49
t(RHO) = B.&7
- 8ASIC - InPOP
InTEMP = -0.7B - 0.0721nMW - 1.10&lnPRIMEUR - 1.24SY + 0.972AFP -
(1.S0)
(S.S6)
(2.BS)
(0.8S)
1.814EFTPP - 0.0042T + 0.0000&4TSO + 0.10102 + 0.2S03 + 0.OS04
3.79)
(2.S&)
(S.66)
(13.02)
(2S.84)
(S.89)
2 • 0.967
F(10; 91) = 272.07
2 = 0.964
DW = 2.043
HO • 0.672
(RHO)=9.lJ
133
5 - BASIC + lnPCWEl
lnTEMP • -1.99 - 0.1461nMW - 0.141nPRIMEUR - 0.521nPOP - 1.29SY +
(4.33)
(10.66)
(2.B7)
(5.03)
1.102AFP - 1.92EFTPP - 0.0013T + 0.000037TSO + 0.OB902 + 0.2303 +
(1.304)
(3.70B)
(0.711)
(3.03)
(13.03
(2B.66)
0.03704 + 0.09771nPCWEl
(5.04)
(5.41)
R2 • 0.96B
F(12; 90) = 217.63
R2 • 0.964
DW = 1.B9
RHO • 0.33
t(RHO) = 3.63
6 - BASIC + EDP
lnTEMP = -1.43 - 0.OB21nMW - 0.1271nPRIMEUR - O.3451nPOP - 1.004SY +
(1.81)
(6.54)
(1.072)
(2.33)
0.227AFP - 1.60EFTPP + 0.003081 + 0.0000156T50 + 0.09202 +
(O.lBB)
(3.36)
(0.B4)
(0.63)
(11.76)
0.23803 + 0.04404 - D.847EDP
(23.68)
(5.19)
(2.2B)
R2 = 0.969
F(lO; 90) = 222.48
R2 = 0.965
DW = 2.05
RHO = 0.648
t(RHO) = 8.63
134
Appendix E:
Teenage Unemployment Equation:
OLS Linear.
BASIC +
PCWEl and BASIC + EDP.
1 - Basic Teenage Unemployment eQuation + PCWEL
THUR ~ -0.116 - 0.00665929MW + 1.61PRIMEUR + 0.343SY + 0.035AfP +
(0.1.5B)
(11.12)
(3.8B)
(0.12B)
0.469POP - 0.202EFTPP + 0.000912T + 0.0000036TSO + 0.02902 +
(0.B19)
(0.14)
(1.83)
(1.003)
(B.34)
0.001303 + 0.003604 + 0.03056PCWEl
(1.91 )
(0.99)
(0.56)
R2 ~ 0.B51
F(12; 91) ~ 45.51
R2 ~ 0.83B
DW" 2.04
135
2 - BASIC Teenage unemployment eguation ~ EDP
TNUR: -0.134 • 0.0230581MW • 1.bObPRIMEUR + 0.414SY - 0.083AFP +
(0.50)
(12.31)
(4.19)
(0.29)
O.b7bPOP - 0.Ob8EFTPP + 0.00132T + 0.00000705TSO • 0.028202 •
(1 .24)
(0.258)
(2.30)
(1.bl)
(1.57)
0.00b103 + 0.0040104 - 0.1705EOP
(1.12)
(1.121)
(1.50)
R2 : 0.8bO
F(12; 91) ·4b.b4
R2 : 0.841
OW : 2.0b
136
Appendix F:
Estimated Impact of an Increase in the MW, FlS and WI on
THUR.
Instrumental Variables Method:
Different
Functional Forms
l
- Teenage unemoloyment eguation.
Simultaneous eguations estimation.
Instrumental variables:
linear form.
TNUR 0 0.561 + 0.0035898~W + 1.049PRIMEUR , 0.3785Y - 0.183AFP -
(0.0855)
(2.41)
(4.06)
(0.565)
0.511POP - 0.235EFTPP + 0.00126T - 0.0000068T5Q + 0.022302 -
(0.572)
(0.896)
(2.29)
(1.46)
(3.906)
0.00080203 - 0.0036204 - 0.J40447FL5 - 0.133848WI
(0.121)
(0.59)
(1.32)
(0.28)
OW 0 2.05
137
2 - Teenage unemployment equationj joint estimation.
GlS instrumental
variables estimation: linear form.
~NUR = 0.327 - 0.003B971BHW + 1.27PRIHEUR + 0.3645Y - 0.156AFP -
(0.096)
(3.B7)
(4.1B)
(0.544)
0.214POP - 0.245EFTPP + 0.0012BT - 0.0000072T50 + 0. 0267 02 +
(0.29)
(0.94)
(2.47)
(l.BO)
(5.35)
0.003903 + 0.000B004 - 0.50B265FL5 + 0.160374WI
(0.70)
(0.15)
(1.29)
(0.6B6)
OW = 1.98
138
3 - Teenage unemployment eguationî joint estimation, l09arlthmic fonm
instrumental variables method.
lnTNUR = 0.386 + 0.03577351nHW + 0.4441nPRIHEUR + 0.7861nPOP +
(0.40)
(4.05)
(1.14)
1.9055Y' 1.545AFP - 0.371EF1PP ,0.00441 - 0.0000108150 '
(1.81)
(0.589)
(0.111)
(1.009)
(0.000)
0.111 401 , 0.007103 ' 0.0390704 , 1 .98011nFL5 , 0.7453JIWI
(5.13)
(1.43)
(0.89)
(0.37)
(0.15)
DW·I.19
139
- Teenage unemployment eguationj joint estimation; logarithmic form.
GLS instrumental variables method.
lnTNUR ~ 0.197 + O.03176971nMW + 0.4571461nPRIMEUR + O.6651nPOP ~
(0.466)
(6.35)
(1.338)
1.775Y + 1.123AFP - 0.971EFTPP + 0.00668T - 0.0000357T50 +
(3.24)
(0.559)
(0.579)
(1.87)
(1.33)
0.22602 + 0.083903 + 0.05204 + 1.539251nFL5 , 1.39704Wl
(6.93)
(2.59)
(1.52)
(0.764)
(0.83)
DW 0
2.098
140
Appendix G:
Extended Teenage Labor Force Participation Equation.
Simultaneous Equations Estimation.
Instrumental
Variables Method.
Oifferent Functional Forms.
Teenage labor force participation equation joint estimation.
Instrumental variables method.
Linear form.
LFP
- 0.2885Y
0
3.86 ~ 0.137554MW ~ 2.97PRIM[UR
- 0.2268AFP -
(2.90)
(6.31)
(2.77)
(0.631 )
4.69POP - 0.187[FTPP + 0.00087T + O. 0000052T50 + 0.036902 +
(4.92)
(0.64 )
(1.402)
(0.987)
(5.72)
0.091803 ~ 0.004904 - 3.16826FL5 + O. 369828WI
(12.48)
(0.72)
(5.16)
(0.69 )
DW = 1.22
141
2 - Teenage labor force participat;o~uation: joint estimation.
Instrumental variables - GLSj linear fann.
TLfP ~ 2.04 - 0.102357MW - 1.42PRIMEUR - 0.355Y • 0.321AfP -
(1.b2)
(4.07)
(2.31)
(0.70)
1.375POP - 0.448EfTPP - 0.001141 '0.000022150.0.05302 •
(1.15)
(2.38)
(0.949)
(2.85)
(12.12)
0.11403 , 0. 013 04
1.42033fL5 , 0.0711244WI
(21.11 )
(2.71)
(2.93)
(0.734)
DW ~ 2.06
RHO ~ D.b15
(RHO) ~ 7.89
142
3 - Teenage labo~ fo~c~~ticipatiQn equation.
Joint estimation.
Instrumental va~iables method. _~~ithmic fo~m.
lnTlFP - -b.18 - 0.1011b31nMW - 0.2J41nPRIMEUR - 1.421nPOP - 2.482AFP
(3.55)
(J.40)
(b.22)
(2.b8)
1.40EFTPP - 0.280595Y , 0.005071
0.0000074150 + 0.05902
(2.19)
(1.15)
(3.25)
(0.00)
(4.11)
0.1003 - 0.02 74 04
6,7'J541nFlS - 0.550290Wl
(9.77)
(1.79)
(0.12)
(0.499)
OW - 1.43
•
143
~ - Teenage labor force participation. Joint estimation.
Instrumenta
variables methodj GLS logarithmic form.
1nTLFP ~ -4.39 - 0.09415181nMW - 0.l911nPRIMlUR - 0.9&11nPOP
(2.14)
(5.91)
(4.020)
l.25&AFP - 0.5105Y - 1.&3lFTPP + 0.002&1T + 0.000011T50 +
(1.2&)
(1.85)
(3.22)
(1.288)
(0.84&)
0.08&302 + 0.19303 + 0.00028204 - 4.&&51nFL5 - 0.04011WI
(1.41)
(13.&2)
(0.022)
(4.17)
(0.122)
OW " 1.94
RHO " 0.359
t(RHO) = 3.88
144
Append1x H:
Results of Covariance Procedure:
Correlation
Coefficients Matrix and Mean, Standard Deviation of TEMP.
FLS, WI, MW, TFLP and THUR.
TEMP
FLS
WI
MW
HFP
THUR
TEMP
FLS
-0.2345
1
WI
0.214
-0.343
1
MW
0.309&
-0.&97
0.491
1
TLFP
0.9&&
-0.429
0.258
0.411
1
THUR
-0.1&4
-0.714
0.14&
0.377
0.094
1
The correlation coefficient expresses the degree of association
between variables.
SA defined, it provides us with a single measure
not only of the direction of the association but the strength of the
relationship between two variables while contro1ing for the effects of
one or more variables.
The correlation matrix above shows a negatlve degree of
association between TEMP and fLS (-O.2345) whereas such degree is
positive between TEMP and MW (0.309b).
This reveals that the
relat1onsh1p between lEMP and MW may be spurious.
145
2 - Mean and standard Deviation
,
Variable
Mean
Standard Deviation
TEMP
0.425769
0.544081 E-01
FlS
0.855251
0.149838[-01
WI
0.014987
0.109215E-01
MW
0.307269
0.913308E 01
TlFP
0.501194
0.633640E-01
THUR
0.150153
0.278693E-01
146
616LIOGRAPHY
Adie. Douglas K.
·Teen-Age Unemployment and Real Federal Minimum
Wage.·
Journal of Political Economy. 81, No. 2. Harch/April 1983.
Adie, Dou~las and Gallaway, L.
l'The Minimum Wage and Teenage
Unemp 1oyment:
A C9Jm1ent. "
Western Economie Journal, December
1973.
Ashenfelter and Heckman.
uThe Estimat 10n of Incarne and Substitution
Effects in a Model of Family Supply."
Econometrica, Vol. 42,
January 1974.
Ashenfelter and R. Smith.
"Compliance with the Minimum Wage Law."
"J~o~u~rn~.~l~o~f~Po~l~i~t~i~c~.~l~Ec~o~n~o~m~y.
July/August 1971.
Betsey L. Charles and Dunson, H. Bruce.
"Feder"al Minimum Wage Laws
and the Employment of Minority Youth. Il
America" Economie
Association Papers and Proceedings, May 1981.
Sowen. William G. and F1negan. T. A.
The Economics of Labor Force
Participation.
Princeton, NJ:
Princeton University Press. 1969.
Brown, Charles; Gilroy Curtis; and Kohen, Andrew.
[ffects of the
Minimum Wage on Youth Employment and Unemployment in Minimum Wage
Study CO""'iss10n, Vol. 5. June 19B1.
Brozen, Yale.
DEffect of Statutory Minimum Wage lncreases on Teenage
Employment."
Journal of Law and Economics, April 1969.
Brozen. Yale.
"Minimum Wage Rates and Household Workers.~
Journal of
Law and Economics, April 1962.
Brozen, Yale and Milton Friedman.
The Minimum Wage:
Who Really
Pays?
Washington, D.C.:
The Free Press Association. 1966.
Cain, Glen.
Married Women in the Labor Force.
Chicago:
The
University of Chicago Press, 1966.
Cox, C. James and Ronald L. Oaxaca.
"Effects of Minimum Wage Policy
on Inflation and Output Prices, Employment and Real Wage Rates by
Industry in Minimum Wage Study Conmission. Vol. 6. June 19B1. p.
171.
Cox, C. James and Ronald L. Oaxaca.
"The Polit1eal Economy of Minimum
Wage Leg;slation.~
Economie Inquiry. Vol. 20, No. 4. October
1962.
147
Ehrenberg. Ronald G. and Marcus, Alan J.
-Minimum Wage Legislation
and the Educat10nal Dutcomes of VouthoN
Research in Labor
Economies. R. Ehrenberg (Ed.), July 1979.
Ehrenberg. Ronald G. and Robert S. Smith.
Modern Labor Economies
Theory and Public Policy.
Scott. Foresman and Company. 1982, pp.
69-79.
Freeman. R. 8.
·Economie Determinants of Geographie and Individual
Variation in the Labor Market Position of Young Persans."
Presented at NBER Conference on Youth Unemployment. March 1979.
Forthcoming in Youth Unemployment:
Us Nature. Causes and
Consequences. edited by R. Freeman and D. Wise.
Freeman. Richard and Ken Clark, "How Elastic is the Demand for
Labor?"
Review of Economies and Statistics, November 1980, p.
518.
Fisher. Alan.
"The Minimum Wage and Teenage Unemployment:
A Comment
on the Literature."
Western Economie Journal, 11. December 1975,
pp. 514-514.
Fortin. Pierre.
"The Priee, Employment and Redistributive Effects of
the Minimum Wage:
Lessons from the Quebec Experience."
Paper
presented at the Conference on Incomes in Canada.
Winnipeg,
Manitoba, May 10-18, 1979.
Frish. Helmut.
-Inflation Theory 1963-1975:
A Second Generation
Survey."
Journal of Economie Literature, 15. Decernber 1977. pp.
1289-1377 .
Gram11ch. M. Edward.
Wimpact of Minimum Wages on Other Wages,
Employment and Family Incornes. 1l
Brookings Pager on Economie
Activity. 1976.
Goldfarb. R.
"The Policy Content of Ouantitative Minimum Wage
Research."
Proceed~ngs of the Industrial Relations Research
Association (IRRA), 1914.
Hamermesh, Daniel.
"Econometrie Studies of Labor Demand and Their
Applications to Public Poliey."
Journal of Human Resourees. Fal1
1976.
Hamermesh. Daniel.
lIMinimum Wages and the Demand for Labor."
Economie Inguiry, Vol. XX, No. 4, Oetober 1982.
Hamennesh, S. and Daniel Grant James.
"Econometrie Studies of
Labor-Labor Substitution and Their Implications for Policy."
The
Journal of Human Resources, XIV, 4, 1979.
148
Heckman. James A.
'Practical Survey of Recent Research on the Labor
Supply of Women."
America" Economie Review. May 1978.
Hughes, James and Richard Perlman.
Economies of Unemployment:
Comparative Study of the UK and the USA. 1984.
Johnson, G. Harry.
"Minimum Wage Laws:
A General Equilibrium
Analysis."
Canadian Journal of Economies, II, No. 4, November
.....
1969.
-.
•
Kaitz, Hyman.
"Experience of the Past:
The National Minimum."
Youth
Unemployment and Minimum Wages~
bulletin 1657, U.S. Oepartment
of Labor. Bureau of Labor Stat;st;cs, 1970.
Katz. Arnold.
"Teenage Employment Effects of State Minimum Wages."
Journal of Huma" Resources, Spring 1973.
Kalacheck. Edward.
"Determinants of Teenage Employment.
Journal of
Human Resources, Winter 1979.
Keely, Michael.
Labor SUDPly and Public Policy:
A Critical Review.
Academic Press, 1981.
Kil'ingsworth, R. Mark.
Labor Supply.
Cambridge University Press,
1983.
Kosters, Marvin, and Welch, Finis.
"The Effects of Minimum Wages on
the Distribution of Changes in Aggregate Employment. u
American
Economie Review, 62, No. 3, June 1972.
Linneman.
"The Economic Impact of Minimum Wage Laws:
A New Look at
an Old Question."
Journal of Political Economy.
February 1982,
pp. 443-469.
Long, Clarence.
The Labor Force Under Changing Income and
Emp loyment.
Princeton, NJ:
Princeton University Press, 1958.
Lovell, Michael.
"The Minimum Wage, Teenage Unemp1oyment, and the
8usiness Cycle."
Western Economic Journal, December 1972.
Lovell, Michael.
"The Minimum Wage Reconsidered."
Western Economic
Journal, December 1973.
Masandri, Hashimoto.
"Minimum Wages and On-the-Job lraining."
American Enterprise Institute Studies. 311, 1981.
MCCulloch, J. Huston.
Macroeconomie Implications of the Minimum Wage
in the Economics of Legal Minimum Wages (edited by Simon
Rottenberg) .
149
Meyer, H. Robert and David A. Wise.
~Discontinuous Distributions and
Missing Persans:
The Minimum Wage and Unemployed Youth.-
Econometrica. Vol. 51, Ho. &, Hovember 1983.
Meyer, H. Robert and David A. W1se.
"The Effects of the Minimum Wage
on the Employment and Earnings of Youth."
Journal of Labor
Economlcs. vol. 1. Ho. 1. January 1983.
Mincer. Jacob.
-Labor Force Participation of Married Women."
Aspects
of Labor Economies, H. G.
Lewis (Ed.j, "Princeton:
Princeton
University Press, 1962.
Mincer, Jacob.
IOUnemployment Effects of Minimum Wages.'l
The Journal
of Polit;ca1 Economy. August 1976.
Minimum Wage Study Commission. Vol. l, V, VII (Report of the),
May/June 1981.
Washington, D.C.:"
U.S. Government Printfng
Office.
Moore, Thomas Gate.
l'The Effect of Minimum Wages on Teenage
Unemp l oyment Rates. l'
The Journal of Political Economy.
July/August 1971.
Morse. C. Laurence.
"Youth Employment:
New Perspectives.
Increasing
Unemployment and Changing Labor Market Expectations Among Black
Male Teenagers."
Ameriean Economie Association Pa pers and
Proceedings. May 1981.
Perlman, Richard R., Kreps. Martin, and Somers.
contemporary Labor
Economics and Labor Relations.
Wadsworth, Inc., 1980.
Perlman, Richard.
Labor Theory.
New York:
Wiley, 1969.
Ragan. James F., Jr.
"Minimum Wages and the Youth Labor Market."
The
Review of Economies and Statistics, Vol.
LIX, May 1977. Number 2.
Ragan. James Jr.
"Minimum Wage Legislation:
Goals and Realities.
Nebraska Journal of Economics and Business. Autumn 197B. 21-2B.
Ragan, James Jr.
"The Effect of a Legal Minimum Wage on the Pay and
Employment of Teenage Students and Non Students."
The Economics
of Legal Minimum Wages, Simon Rottenberq (Ed.), 1981. AEI.
Reder. Melvin.
"The Theory of Occupational Wage Oifferential5."
~erican_Economic Review, VXV, Oecember 1955.
Rottenberg. Simon.
"The Economlcs of Legal Minimum Wages.
1981
Amerlcdn Enterprise Institute for Public Policy Research,
Washington.
150
Sargan, J. O.
·A Model of Wage-Price Inflation.-
Review of Economie
Studies. 1980. XLVII. 97-112.
Se11ekaerts. Brigitte.
Mlmpact of Minimum Wage Legislation on Wage
and Priee Inflation. N
Minimum Wage Study Commission. Vol. 6,
June 1981.
Silk, Leonard.
"Oealing with Joblessness. " New York Times,
Wednesday. Oecember 8. 1982.
Singer, W. James.
liA Subminimum wage - Jobs for Youths or a Break for
their Employers?"
National Journal, January 24. 1981.
stein, Robert L.
-New Definitions for Employment and Unemployment."
Employment and Earnings, February. 1967.
Stigler, George J.
"The [conomies of Minimum Wage Legislation.'1
America" Economie Rev;ew. June '9~6. pp. 358-365.
SUl1Il\\ers. Lawrence.
MMeasuri ng Unemp l oyment."
8rook i n9 Paper on
Economie Activitv. No. 2, 1981.
Sweet. A. James.
Women in the Labor Force.
New York:
Seminar Press.
Inc., 1973.
Swid1nsky, Robert.
MMinimum Wages and Teenage Unemployment."
Canadian Journal of Economics, XIII. No. 1. February 1980.
U.5. Department of Labor.
MMinlmum Wage and Maximum Hours Standards
Under the Fair Labor Standards Act Submitted to Congress. 1970.
Welch. Finis.
"Minimum Wage Legislation in the United States."
[ ~c,,0,,0o",m"l,,'c,--,l"n"g"u"i-'.r.:t.Y. Jan ua ry l 977 .
Welch. Finls.
"Minimum Wage Legislation in the Unîted States:
Reply .•
Economie lnquiry. January 1977.
Welch. Fini'.;.
"Minimum Wage:
Issues and Evidence."
American
Enterprise Institute Studies, 206. 1978.
Welch. Finis and Cunningham. James.
"Effects of Minimum Wages on the
Level and Age Composition of Youth Employment."
Revie..., of
Economics and Statistics, February 1978.
Welch, Finis and Kosters, Marvin.
"The Effect::. of Minimum Wages on
the Distribution of Changes in Aqqregate Employment."
Americ~n
Economic Review. June 1972.
Wessels, Walter J.
"The Effect of Minimum Wages in the Presence of
Fringe Benefits:
An Expanded Model. "
Eco~Qmic lnguiry. 18,
1980, 293-313.
151
West, E. 6., and McKee. Michael.
"Monopsonyand 'Shock' Arguments for
Minimum Wages.-
Southern Economie Journal. January 1980.
Zucker. Albert.
~Minimum Wage and the Oemand for Low-Wage Labor. M
Quarterly Journal of Economies. May 1973.
RESUME DE LA TRESE DE Ph.D.
SN.JiOJNOMIE DE ALLEOU M' BEr
Titre
'Mininrum Wage, lruflation and Unemployment
a 5 imul taneous
equations analysisll •
ou
Il
Salaire Minimum, l'Inflation ct le chômage
une analyse
à équations simultanées"
Présentée à llUniversité
du Wisconsin - Milwaùcee V.S.A
le 14 Décembre 1984
-------_._------_._--_.._------++++++++++++++++------------------------------
C('t 1.,' 1 hèsc analyse les effets du Salaire Minimum Fédéral sur
l 'l:ll\\ploi \\..lU ::"Hl levers, le chôm..'-ige, aux USA; plus particulièrement sur
le chômage dcs Jdolescents (16 - 19 ans),
&1 effet, plus de quarante ans après sa ratification, la loi
du salaire Minimum demeure encore controversée. Des recherches antérieures
ont discuté de cette controverse en exam~t divers effets sur l'emploi
ou sur le chômage du relèvement du salaire m:inimtDTl au fil des années.
Concernant les effets du salaire minimtDTl sur l'emploi J les
études les plus nkmltes ont abouti à des conclusions conflictuelles .
.. .; ...
- z -
D'une part il a été reconnu que la g~l~ire minimum a un impact
substantiel sur l'emploi, spécialement sur l'emploi des adolescents.
D'autre part, il a été démontTé que l'effet dll sallllre minimum s'évanouit
dès lors que l'on tient compte de la composUjon changeante de la popu-
lation active. L'évidence collective de ces recherches Tend largement
compte des différents effets du salaire minimum sur différents groupes
sociaux.
Mais quelle
que soit la nature ~lriée des résultats, presque
toutes ces études utilisent une méthodologie camnlln1e. Celle-ci consiste
en une équation unique dans laquelle lIon regIE'sse le salaire minimum
et d'autres variables pertinentes sur me mc~,ure du la population active
(ou de la force de travail). Cependant, certainees considération:, ren-
dent cette approche suspecte.
D'abord, l'LU1 des effets du salaire minimum largement
admis
est l'augmentation du coût du facteur travail. Les en~loyeurs pour tenir
compte de ce coût élevé, réagissent en fixant. des prix élevés à leurs
produits ; ainsi il y a me génération simllltéuI6e de la spirale salaire-
inflation et prix-inflation.
Ensuite, l'l.Dl des phénomènes les plus remarquables dans l'his-
toire de la population active américaine est la croissance séculaire du
taux de participation des femmes dans l'activité économique et plus par-
ticulièrement des femmes mariées.
Néanmoins, la plupart des travaux antérieurs ont manqué d'in-
corporer de manière explicite ces facteurs dans leurs modèles d'analyse
et les résultats obtenus ont dû être 'biaisés •
.../ ...
- 3 -
Pour éliminer ceS biais, cette dis.s(~rtat Ion développe lm
modèle de trois équations simultanées qui comprend l'uffre de travail
féminine et l'inflation générée par les coûts du fncteuT travail.
L'objectif
principal de cette recherche cst donc d 1 apporter
la ltunière SUT cette controverse du salaire minirll1m fédéral en employant
une méthode nouvelle et en y incluant des factoU1'S nouveaux qui sont ap-
parus SUT le marché du travail.
Afin de rendre nos résultats comparables à ceux de travaux
antérieurs, cette dissertation a utilisé des ,k)nnées trimestrielles de
1954 à 1979.
Ces données furent collectées de ttCli9 Sources principales
la "eurrent Population Sur/ey", "Ernployrnent and Eamings" et la base
de daMées sur fichier de la Citibank appelée "Citibase".
Vestimation empirique s'est faite pur étapes. D'abord. les
variables relatives à l'offre de travail des J:l!n~lelt ct celle de l'infla-
tion par les coûts furent introduites comme ,~ogènes.
Mais puisque ces
deux variables sont eUes-mêmes déterminées pltr d'Hutres facteurs dont
le salaire minimum entre autres~ elles furent onsuite introduites comme
variables endogènes et dès lors la méthode dl t'!quation simultanées a été
employée pour évaluer l'impact de ces deux vilriables sur 11 emploi (ou sur
le chômage) des adolescents. La technique des moindres carrés à deux étages
et ceUe des variables instnunentales furent employées, alternativement.
Concernant l'introductio:1 des nouveUe v(lriables explicatives
connne exogènes, les résultats démontrent une très grande sensibilité
entre la méthode des moindres carrés ordinain;'~ et celle des moindres
carrés généralisés.
Mais â cause de la présence d1autocorrelation, l'ac-
cent est porté sur les résulLats obtenus n r>lU'1 iT dcs mQinùreCj carrés
généralisés.
. .. / ...
•
•
- 4 -
Les estimations passées révèlent qui une aUgfl1cntation d" 1(~
pour cent du salaire mini:nurn cOi1duisait à une rédul.:t JOI1 de 1 ~ 3 pour
cent du niveau d'emploi des adolescents. Nos r(isllHats montrent tme
réduction de ZéTO à Lm pOUT cent de
réduction du nivc<lu d'emploi des
adolescents suite à 10 pour cent d'augmentatlon du salaire minimum.
Mais dès que lIon étend le modèle aux dCl~ nouvelles variables,
l'on note des variations considérables dans les résultats. En substance,
l'effet négatif du salaire rnininnM disparaît presque totalement ; et dans
tous les cas, cet effet n!est plus significatif statisquement.
En revanche la variable "offre de travail fém:1Jline" exhibe
un effet négatif très fort et statistiquement significatif sur le niveau
d'emploi des adolescents. Nos estimations montrent qu'une augmentation
de 10 pour cent de l'offre de travail féminine concluit à Lme reduction
d1environ 28 pour cent du niveau d'emploi des jeunes. Et cet effet
négatif s'est maintenu à travers différente spécification et diffférentes
formes fonctionnelles, linéaire et log-linéaires.
Quant à la variable "inflation pat lets salaires", elle montre
lD1 effet positif sur l'emploi des jeWles) mais cet effet n'est pas
significatif •
Nous avons égalenent exploré le con~ortement des variables
e},:plicatives en utilisant un système d'équations simultanées quaI!d ces
variables sont endogènes. Ici encore, l'on découvre des variations
importantes au niveau des résultats.
Le salaire minimwn a lD1 effet negatif très réduit et non
significatif s'Jr l'emploi des adolescents. 1a vln,j,ahle "inflation des
salaires" montl-c quant à elle Lm effet positif
mais toujours non SlgTU-
ficatif. Seule la variable de l'offre de travail des f(.·mmes démontre
avec consistence tm effet négatif important et ~itatlsquement signlficatif,
.../ ...
- 5 -
quelle que soit la spécification et la forme TI)nctionnelle utilisée.
Nos estimations montrent qU'à la suite d'une il11~r.Hlen[<Jtjon de 10 pour
cent de salaire minimum et de l'07-frC' df' tr<l~'~lll
r"llinlrlc,
il s',:nsuIT
une réduction de 0,34 pour cent ('t de 2(1,83 pOli) l'l'Ilt: d\\l niveau d'emplol
des adolescent<=., res.pectiverr,cnt. Ces variilbLc,; (1IIl: le mÔlIl' effet quand
la variable à expliquer est le taux de partiCJ.Jl~t il111 de la population
active des adolescents. Nous aVUllS t muvé J COl1unc cl 1autres études de
par le passé, un retrait des adolescents de ü( population active à la
suite d'une augmentation de la population acUve. Néanmoins, le retrait
de la fOTce de travail émanant d'une augmentat ion de l'offre de travail
des femmes est ù'une grandeur plus élevée.
Les études antérieures ont accepté 10 retrait étormanunent grand
des adolescents de la population active sans autre explication que celle
imputable à l'effet négatif du salaire minirmml. Mals l'on note que l' in-
flux des travailleurs de sexe féminin dans la forco de travail s'est
accru d'une proportion de l.D1 tiers avant la E~('ondc guerre mondiale à
Son niveau actuel de cinquante trois pour cent. L'on peut donc faire
l'hypothèse légitbne que cette croissance de la force de travail féminine
est un facteur sous-jacent à prendre en compte dans 1'6valuation des
déterminants du retrait des adolescents de la force de travail.
Bien sOr, cette hypothèse n'indiqUe fias nécessairement une
relation de causalité entre 10 taux de participation des i;1dolescents à
la force de travail, et l'offre de travail fôudnine II lu hausse. Plutôt
nos résultats sont cLlnsistents avec la forte corrélation négative entre
ces deux groupes de travailleurs qui peut expl.:1quer "à quel degré les
femmes,surtout les femmes mariées et les adolescents sont des facteurs
substituables dans la P:-oduction" cormne l'ont observé Hamennesh ct Grant.
... / ...
•
- 6 -
Les résultats deviennent difficiles 9 interpréter quand le
tauoc de chômage des adolescents est utilisé c~lm~ la variable dépendante.
Les effets d'une croissance des variables expliC'.atives concernées sur le
taux de chômage sont soit très petits, soit du signe contraire.
En outre, les coefficients estimés e:V\\ibent tme très, grande
imprécision comme le dênnntrent les ècart-types des flt'tOurs très élevés.
Cette imprécision et cette insignifLutce dEIS rc.1sultats accordent
du crédit à la notion déjà acceptée selon laquelle 10 taUX d'emploi et le
taw:: de participation à la force de travail, sont de meilleures mesures
descriptives des conditions du marché de travail que ne l'est le taux de
chômage qui est entâché de problèmes d'erreurs de compta~e.
Les implications de fannulation de politique d'emploi des résul-
tats de cette recherche sont de grande importllnC(~. En effet, l'un des
projets souvent suggerés pour résorber le tau;( <ok chôlllage élevé de~; adoles-
cents consiste en la fixation d'ill\\ "sous-salaire ,dnimurn" ou "d'Lm salaire
-_._---_._...•. ..
~
minimum
dual Il
pour les adolescents.
Les débats au congrès des USA sous la pous~ée de l' administrat ion
Reagan en est Lme expression. Nos résultats suggèrent qu'une telle pOlitique
de réductiOn de salaire minimum à Lm taux inférieur spécifiquement pour les
adolescents, si elle est appliquée isolement J ne saurait être le remède
espéré contre le chômage global des adolescents. Plus important en~ore, les
résul tats de nos recherches mettent en doute l' ad~quntlon cl 'tm.e proposition
de sous-salaire minimum comme une politique viable pour résoudre le taux
de chômage élevé des adolescents.
D'un côté, les résultats de nos recherches ont rnontr6 qu'une
augmentation dans la variable offre de tra~il des femmes a eu de manière
consistente, un effet suffisamment négatif et statistiquement significatif
... / ...
•
7
sur le niveau dtemploi et sur le niveau de participation à la force de
travail des adolescents. De l'autre, le salaire minimum a revélé une
certaine ambiguité quant à son effet néglltif à la fois sur le niveau
d'emploi et sur le niveau de leur participation ~ la force de travail des
même adolescents, surtout quand le modèle est e!rtullé en utilisant le sys-
tème d'équations sumultar.ées.
La variable de l r .lnflntion par les salaires
a eu constamment un effet positif SUT l'emploi et la participation des
adolescents; mais cet effet est si minime que l'on peut avancer sans
grand risque qu'elle n'a pas d'effet significat:Lf SUT les variables à
exp liquer •
En conséquence, étant donné
l'incertitude entourant l'effet
d'une augmentation de salaire minimum sur l'eml)10i et le taux de partici-
pation des jeunes J l'on devrait considérer la pl'oposition de la "differen-
tielle" du salaire minimum avec attention. A llh nivenu conceptuel, le
"salaire minimum dual" DU "salaire
minimum diê~ére~l·.iéH, selon ses avocats
défenseurs. réduirait le taux de salaire que lC~1 employeurs paieraient aux
adolescents en dessous du niveau national du shltlhe lninJmum et toujours
se confonner à la loi du "Fair Labor Standards.}ct.ol ùe 1938 qui a établi
le salaire minimum au plan national. Ainsi donc, avec un salaire plus bas.
continuent les défenseurs, ce "salaire minimun dual" réduirait le chômage
des adolescents car nombreux d'entre eux seraient employés selon les pres-
criptions de l'économie de travail classique. (~ncrètement il a été suggeré
au congrès que l'on paie aux adolescents environ 75 li 85 pour cent du
salaire minimum.
En revanche, les opposants soutiennent qu'un tel salaire minimum
dual sera établi aux dépens des anciens travaillE'urs HU taux de salaire
élevé ; car les employeurs soucieux de réduire les coOts de Production et
de maximiser le ?rofit, vont subtituer les jeunes aux vieux. Cela poserait
d'autres problèmes sociaux graves car les travailleurs âgés ont des respon-
sabilités familiales et leur mise au chômage par substi tut ion aura des
conséquences sérieuses. Mais comme cela a été mentionné plus haut, l'on
... / ...
- a -
ne peut ignorer l'influx énonne des feJlD'nC5 5\\11' le marché du travail si
lion veut offrir une analyse complète de lleffet du salaire minimum.
En effet, si ce salaire minimum dual- proposé doit aider à
réduire le chômage des jeunes comme lion antlcipe, il doit être établi
en dessous de cc que l'on appelle le taux de salaire de réservation des
fenmes c'est-à-dire un "taux frontière" quiÛ'\\cite la femme à choisir
entre le Iltravail de foy~rl1 et le travail réJl'Il..,dré l S~ et seulement si
le salaire minimum est en dessous du
taux. de l'~servatlon des ferrones J
alors cela pourrait encourager les femmes à se mettre en retrait du
marché et induire les jeunes à occuper les emplois vacants ou potentiels.
Mais le problème est qu'à ce niveau là, le salaire minimum sera si bas
que seuls les travailleurs immigrés sans aucun taux de résenration,
accepteraient de travailler.
En plus, les adolescents ne constituent ;Jas un groupe homogène
faisant face à un taux de salaire minimum urliC\\~le. l'nT exel'tple l'on a
observé (page 20 de la thèse) qu'en 1978, un té!Ux ,le chôrrnge de 13.9 pOLir
cent pour les adolescents blancs contre 36,3 pour (ent pOUT les <ldoles-
cents noirs.
Si donc d'autres facteurs tels que ] a cliHcl'inlination raciale
mterviennent dans la détennination du ch6lTU1ge, alors lIne politique
s~le d'un taux de salaire dual ne saurait être effective.
Nous avons donc suggeré
au commencement que nous serons en
faveur d'une politique économique conduisant ~ une croissance globale
continue qui permettrait à l'économie d'aborder la force de travail en
augmentation, plutôt que d'adopter une optirnl politlque dont l'effet
ultime serait de confronter un segment du Matché da Travail à un autre.
Dans une économie en expansion, le taux de salaire sera basé SUT le
nombre d'emplois et non sur les caractéristi.ques des gens qui occupent'
ces emplois.