TESTS OF MARKET POWER IN IRISH MANUFACTURING INDUSTRIES
By
G.E. BOYLE
DEPARTMENT OF ECONOMICS,
NATIONAL UNIVERSITY OF IRELAND, MAYNOOTH
ABSTRACT
I ’s been a long-s anding assump ion ha I ish manu ac u ing i ms a e p ice ake s in
hei ou pu ma ke s. This assump ion has been alida ed by se e al agg ega e le el
s udies. While a much smalle numbe o s udies ha e examined his issue a a mo e
disagg ega ed le el, hey end o suppo his conclusion. All o hese s udies also assume
ha i ms a e p ice ake s in hei inpu ma ke s, o , in o he wo ds inpu s a e pe ec ly
elas ically supplied o i ms. This assump ion, howe e , has ne e been o mally es ed.
The e is in insic in e es in he con ex o compe i ion law in asce aining e idence o
de ia ions om pe ec compe i ion in sales o p oduc and pu chases o inpu s. Mo eo e
he in oduc ion o minimum wages p o ides an impo an addi ional mo i a ion o he
opic o his pape , since i is well known ha he in oduc ion o minimum wages o
inc eases in he le el o minimum wages can lead o inc eases in employmen o e a ce ain
ange o wages i he i m possesses ma ke powe in i s labou ma ke . By using ou -digi
le el Census o Indus ial P oduc ion (CIP) panel da a he pape se s ou o es he ex en
o po en ial ma ke powe in I ish manu ac u ing indus ies. The pape employs he
ingenious me hod p oposed by Hall and la e modi ied by Roege in his exe cise. While
he Hall-Roege me hod was o iginally conce ned wi h impe ec compe i ion in ou pu
ma ke s, i can be eadily ex ended o inpu ma ke s. The empi ical esul s do no indica e
much e idence o signi ican impe ec compe i ion in ou pu ma ke s bu he esul s do
poin o e idence o ma ke powe in ce ain inpu ma ke s and in some indus ial sec o s.
The implica ions o hese indings a e discussed.
KEY WORDS: ma ke powe , impe ec compe i ion, minimum wages, an i- us
legisla ion, econome ic es s.
JEL CODES: D21, D43, J42,l4, l60.
TESTS OF MARKET POWER IN IRISH MANUFACTURING INDUSTRIES1
1. In oduc ion
This pape applies a es o ma ke powe , o iginally p oposed by Hall (1988) and subsequen ly
modi ied by Roege (1995), o I ish manu ac u ing indus ies using he Census o Indus ial
P oduc ion (CIP) da abase o he pe iod 1991 o 1999. Bo h Hall and Roege we e in e es ed in
monopoly powe bu hei app oach can be easily ex ended o he case o monopsony powe as
demons a ed in an in e es ing applica ion o he US obacco indus y by Rape , Lo e and
Shumway (1998). Apa om he in insic in e es in he con ex o compe i ion law in
asce aining e idence o de ia ions om pe ec compe i ion in sales o p oduc and pu chases o
inpu s, he in oduc ion o minimum wages p o ides an impo an addi ional mo i a ion o he
opic o his pape . I is well known ha he sec o al employmen esponse o he in oduc ion o
a minimum wage will no be nega i e and may well be posi i e i he sec o is cha ac e is ic by
monopsonis ic- ype beha iou in he labou ma ke (see o example, Boyle, McCa hy and
O’Neill (1998)).
A succession o pape s ha e demons a ed con incingly ha p ice- aking beha iou is a
easonable assump ion as a as ou pu ma ke s a e conce ned o he I ish manu ac u ing sec o
as a whole. A smalle numbe o pape s ha e e i ied he alidi y o his assump ion o
pa icula sec o s o manu ac u ing indus y (see, o example, Callan and Fi zGe ald (1998)). To
he bes o ou knowledge no pape , howe e , has looked a whe he i ms a e p ice ake s in
hei inpu ma ke s. A u he aim o his pape is o a emp o ill his gap.
The pape is o ganised as ollows. Sec ion 2 ou lines he es p ocedu e while Sec ion 3
discusses he da abase employed in he empi ical analysis and p esen s a numbe o econome ic
es s o he p esence o impe ec compe i ion in ou pu and inpu ma ke s. Sec ion 4 concludes
he pape .
1 I would like o hank Denis Conni ee, Donal O’Neill and Mau ice Roche, oge he wi h colleagues a a
depa men al semina , o commen s on an ea lie d a .
2
2. Tes s o Ma ke Powe – he Hall and Roege App oach
Hall (1988) se s ou a new es o monopoly powe in U.S. indus y. He ejec s he adi ional
conjec u al a ia ions’ app oach2 o es ing o ma ke powe , as being o e ly dependen in i s
execu ion on speci ic unc ional o m assump ions, in a ou o a non-s uc u al educed- o m
app oach. His basic bu powe ul insigh is ha he adi ional Solow esidual (SR) should be
independen o a ia ions in he log-change o ou pu in he absence o monopoly powe .
Suppose we ha e a cons an - e u ns wo-inpu p oduc ion unc ion. Equa ion (1) in e p e s he
adi ional Solow esidual as he sum o (unobse able) p oduc i i y shocks (A) and Hicks-
neu al echnical change (

).

 AxxsxySR )()( 12
*
21 (1)
whe e,

y = (log) g oss ou pu ,

x’s = (log) inpu s,
*
2
s = cos sha e o inpu x2 in he alue o ou pu ,

A = (log) p oduc i i y shock,

= (log) Hicks-neu al echnical change.
To illus a e how he Hall es o monopoly powe wo ks we i s de ine he Le ne coe icien
(

mp) o monopoly powe as
mp
mc
mcp


 (2)
whe e,
3
p = ou pu p ice
mc = ma ginal p oduc ion cos
The ‘ ue’ cos sha e ( ) unde compe i i e p icing may be w i en
*
2
s
2
2222
*
2)1(
)()( s
p
p
ymc
xw
ymc
xw
smp

 (2)(a)
and i , hen . 0
mp

2
*
2ss 
Subs i u ing his exp ession o in (1) we can ob ain wo equi alen in e p e a ions
*
2
s3 o he
Solow esidual. Fi s we ha e
mpmp
mp A
xyxxsxySR








 1
)(
1
)()( 11221 (3)
and secondly we ge

 AxxsxxsxySR mp )()()( 1221221 (3)(a)
The addi ion o an e o e m o equa ions (3) and (3)(a), which also inco po a es he
unobse able p oduc i i y a iable

A, allows he Le ne coe icien

mp o be es ima ed using
econome ic me hods.
Rape , Lo e and Shumway (1998) demons a e ha Hall’s app oach can easily be ex ended o
he monopsony case. In he case o monopsony, he Le ne coe icien becomes
2 Fo a ela i ely ecen applica ion in his gen e see Sch oe e (1988) and o an o e iew o
some o he app oaches o es ing o ma ke powe see, o example, Massey (2000)
3 I would like o hank Denis Conni e o d awing my a en ion o his poin which is no made by Hall o indeed, o
he bes o my knowledge, by any o he au ho in he ele an li e a u e.
4
ms
w
w mp


 (4)
I pu chases o say x2 in equa ion (1) we e subjec o monopsonis ic powe hen he cos sha e
e m becomes s2(1+

ms). Inco po a ing his exp ession o in (1) p o ides he ollowing
equa ion o he Solow esidual
*
2
s

 AxxsxxsxySR ms )()()( 1221221 (5)
Again he addi ion o an e o e m o (5) ha inco po a es he unobse able p oduc i i y
a iable, allows

ms o be es ima ed using eg ession analysis.
An impo an implica ion o (5) is ha i is seen o be iden ical o (3)(a). Thus o he wo-inpu
case i is clea ha i is impossible by using Hall’s me hodology o disc imina e be ween
monopolis ic and monopsonis ic sou ces o impe ec compe i ion. Hall acknowledges as much
when he sugges s ha a possible explana ion o his es ima e o a ela i ely la ge and s a is ically
signi ican Le ne coe icien , ha is ob ained by es ima ing an equa ion like (3), could be
monopsonis ic- ype beha iou . Thus a bes i is misleading o e e o es s such as (3) o (5) as
es s o “monopoly” o “monopsony” powe .
A mo e gene al speci ica ion would be o allow o bo h monopoly and monopsony sou ces o
ma ke powe and o es o ei he o bo h sou ces using app op ia e es ic ions. Suppose hen
ha we pe mi ed (1) o be a lic ed by bo h monopoly and monopsony. Wi h his mo e gene al
speci ica ion equa ion (2)(a) becomes
2
2
2
2
2222
*
2)1)(1(
)()( s
w
w
w
mp
p
p
ymc
xw
ymc
xw
smsmp

 (6)
Subs i u ing (6) o in (1) we could ob ain exp essions ha a e iden ical o equa ions (3) and
(5) excep ha he coe icien o be es ima ed is he p oduc o he monopoly and monopsony
Le ne coe icien s. A hi d o mula ion can also be easily de i ed as
*
2
s
5

mp
ms
mp
mp A
xxsxySR









1
)()(
11221 (7)
A i s glance, i would appea possible o use (7) o simul aneously es o impe ec
compe i ion in bo h ou pu and inpu p ices. Howe e , on close inspec ion he igh -hand-side
a iables a e seen o be nea ly pe ec ly collinea .
Hall acknowledges ha he econome ic es ima ion o equa ions (3), (5) o indeed (7) is
p oblema ic because o he inhe en co ela ion be ween he igh -hand-side a iable and he
e o e m which i will be ecalled includes in e alia he e m,

A. In p inciple ins umen al
a iable es ima ion can be employed bu i will be di icul o ob ain sui able ins umen s wi h
he desi able s a is ical p ope ies.
Roege ’s (1995) con ibu ion s ems om his conce n wi h Hall’s o he wise ingenious me hod.
Roege i s o all poin s ou ha he Solow esidual can be equi alen ly ob ained om he cos
unc ion which is o cou se he dual o he p oduc ion unc ion. Deno ing he esidual in his
ins ance by SRP, i s alue is gi en by

 AwwswpSRP )()( 12
*
21 (8)
Roege ’s basic con ibu ion is o no e ha apa om a andom e o e m ha cap u es
measu emen e o , he di e ence be ween he quan i y and p ice-based Solow esiduals should
anish unde he null hypo hesis o pe ec compe i ion in p oduc and inpu ma ke s. Thus we
ha e
0)]()[()]()[( 1122
*
211  wxwxswxpySRPSR (9)
6
I p oduc ma ke s we e, howe e , cha ac e ised by monopoly beha iou hen eplacing in (9)
by he exp ession in (2)(a), he ollowing equa ion p o ides an econome ic es
*
2
s
4





 )]()[(
111 wxpySRPSR mp
mp (10)
whe e,

= andom e o e m.
The g ea ad an age o Roege ’s app oach is o cou se ha since

can be in e p e ed o only
cap u e measu emen e o he e is no eason o suppose ha he igh -hand-side a iable will be
co ela ed wi h i , implying he OLS will yield a consis en es ima e o .

mp.
Simila ly, i monopsony beha iou applies o pu chases o , o ins ance, inpu x2, hen an
econome ic es o i s p esence is gi en by equa ion (11)

 )]()[( 11222 wxwxsSRPSR ms (11)
A u he p ac ical ad an age and a ac i e ea u e o Roege ’s app oach wo h no ing is ha
expendi u e da a a e ypically mo e eadily a ailable han quan i y in o ma ion, especially in
ega d o capi al s ock da a.
Howe e , ou obse a ions on he Hall me hod conce ning he di icul y o disc imina ing
be ween monopolis ic and monopsonis ic sou ces o impe ec compe i ion apply equally o
Roege ’s me hod. Thus es ima ion o (10) o (11) will p o ide essen ially he same in o ma ion
on he ex en o depa u es om pe ec compe i ion. Es ima ion o equa ions o he o m o (11),
howe e , p o ide us wi h a somewha non- igo ous, i less han comp ehensi e, means o
disc imina ing be ween monopolis ic and monopsonis ic sou ces o impe ec compe i ion. I we
allow o a i m o exe cise po en ial ma ke powe in bo h ou pu and in up o M-1 inpu
ma ke s, we subs i u e an exp ession o he o m (6) in (9) o yield
4 This is he dual coun e pa o equa ion (3). We can also de i e an exp ession simila o equa ion 3(a).
7

 

)]()[()( 11
2
wxwxsSRPSR jjj
mpms
j
ms
j
mp
M
j
(12)
o , w i ing (12) using simple no a ion

 

j
M
jjzy
2
(13)
whe e,
)]()[(
and
11 wxwxsz
SR
P
SR
y
jjjj
mpms
j
ms
j
mp
j




The equa ion in (13) implies ha he non- ejec ion o a es o equali y o he

js would s ongly
sugges ma ke powe in he se ing o ou pu p ices whe eas ejec ion o equali y o he

js
would p o ide easonably s ong suppo o a monopsony basis o depa u es om impe ec
compe i ion. The es is o cou se p edica ed on he p esump ion ha equali y o he monopsony-
based Le ne coe icien s o all M-1 inpu s is a highly unlikely p ospec .
3. Applying Roege ’s Me hod o I ish Census o P oduc ion Da a
3.1 Da a
The I ish Census o P oduc ion (CIP) da abase as published by he Cen al S a is ics O ice
(CSO) is employed o es o he p esence o po en ial ma ke powe in he majo i y o
manu ac u ing indus y sec o s. Fo each yea he ull da ase p o ides in o ma ion on some 138
ou -digi indus ial sec o s. Fo each yea da a a e a ailable on he nominal alue o each
sec o ’s g oss ou pu oge he wi h each sec o ’s expendi u e on “Ma e ials”, “Se ices”, “Fuel
8
and Powe ”, “Indus ial Employmen ” and “O he Employmen ”. Unde he assump ion o CRS,
capi al cos s a e implici ly gi en as “ he emainde o ne ou pu ”. A he commencemen o his
s udy we had access o nine yea s o such da a om 1991 o 1999.
The da ase allowed us o o m a panel. In o ming he panel we had o ensu e ha he same ou -
digi sec o agg ega es we e a ailable each yea . As we also wan ed o gene a e sepa a e se s o
ma ke -powe coe icien es ima es a he wo-digi le el, we also con ined he analysis o hose
wo-digi sec o s o which we had a su icien panel o da a poin s o conduc a sepa a e
eg ession analysis o each wo-digi sec o . These conside a ions esul ed in 109 ou -digi
le el obse a ions ac oss 17 wo-digi sec o s being a ailable o each yea . As we lose one
yea ’s da a due o he eg ession a iables being de ined in e ms o log changes, he o al
numbe o obse a ions in he panel is 8x109= 872 o he pe iod 1992 o 1999.
In Table 1 we u nish he lis o wo-digi sec o s o which ma ke -powe coe icien s we e
es ima ed oge he wi h selec ed cha ac e is ics o hese sec o s. Ou chosen sec o s accoun o
nea ly 84% o bo h o al manu ac u ing employmen and “Wages and Sala ies” expenses. The
“Food” sec o (code 15) s ands ou as he single-la ges indus ial sec o in ou analysis. I is seen
o accoun o nea ly 20% o o al manu ac u ing employmen and “Wages and Sala ies” and
a ound a qua e o o al “Ma e ials” and “Se ices” expenses. Sec o 24 (“Chemicals”) also
s ands ou as a sec o o signi icance. I is clea ly a capi al-in ensi e sec o and accoun s o
nea ly 40% o “Capi al” expenses o all manu ac u ing indus ies. Fo abou six5 o hese
sec o s a leas hal o he g oss ou pu is p oduced by o eign-owned i ms and o e hal 6 o he
sec o s expo a leas 60% o he ou pu ha is p oduced.
5 Sec o s 17-18, 22, 24, 31-33 and 34-35.
6 Sec o s 15, 17-18, 22, 24, 25, 28, 31-33, 34-35 and 36-37.
9
Consis en wi h ou indings in Table 3, we ind a p eponde ance o signi ican ma ke -powe
coe icien es ima es o he “Ma e ials” (13 sec o s) and “Se ices” (11 sec o s) inpu s. Abou
se en sec o s epo s a is ically signi ican alues o “Indus ial Employmen ” wi h i e and
ou sec o s e u ning s a is ically signi ican alues o “O he Employmen ” and “Fuel and
Powe ” espec i ely.
Abou eigh sec o s epo a ma ke -powe coe icien o “Ma e ials” ha is g ea e han o equal
o 0.5% (sec o s 15, 17, 18, 21, 24, 26, 32 and 36). The su p isingly nega i e and la ge ma ke -
powe coe icien ha was ob ained o “Se ices” in Table 3 is eplica ed o i ually e e y
sec o in Table 4. Abou se en sec o s epo a ma ke -powe coe icien o “Se ices” ha is
g ea e han o equal o –5% (sec o s 17, 20, 24, 26, 28, 32 and 33). A ound eigh sec o s (sec o s
15, 17, 20, 26, 28, 34, 35 and 36) e u n an es ima ed ma ke -powe coe icien in espec o
“Indus ial Employmen ” ha exceeds o equals 1%.
Th ee sec o s s and ou as ha ing especially la ge and s a is ically signi ican ma ke -powe
coe icien s in espec o “Ma e ials”, “Se ices” and “Indus ial Employmen ”, namely, sec o s
17 (“Tex iles”), 24 (“Chemicals”) and 28 (“Fab ica ed Me als”).
3.2.3 Some obus ness checks: in luen ial obse a ions
Ou esul s so a sugges he p e alence o ma ke powe in he p icing o p oduc ion inpu s in a
la ge numbe o I ish manu ac u ing sec o s. Gi en he impo ance o hese indings i is
impo an ha ou esul s a e subjec o a se ies o obus ness’ checks. Ou i s se o es s is
mo i a ed by he conce n ha ou OLS es ima es may be unduly in luenced by ou lie
obse a ions. The possibili y o ou lie s is p omp ed by he e y na u e o he me hodology
employed o gene a e he esul s in Tables 3 and 4. I he hypo hesis o pe ec compe i ion in
ou pu and inpu p ices holds o a la ge numbe o obse a ions hen acco ding o equa ion (9)
he dependen a iable akes on he alue o ze o. Whe e impe ec compe i ion holds hen (9)
akes on non-ze o alues. I is possible he e o e ha hose obse a ions which ha e non-ze o
alues o he dependen a iable will unduly in luence he eg ession esul s.
We employed wo app oaches o assess he impac o excep ional obse a ions. Fi s , we
pe o med OLS on a unca ed da ase which excluded ou lie obse a ions acco ding o he
16

c i e ion p oposed by Belsey, Kuh and Welsch (1980)7. In Table 5 we epo he es ima es o
ma ke powe ha we e ob ained on he assump ion o cons an wo-digi sec o slopes. Abou 90
o he 872 obse a ions we e deemed o be “in luen ial” acco ding o he Belsey-Kuh-Welsch
es . These esul s should be compa ed di ec ly wi h hose gi en in Table 3. These es ima es
se e o highligh he quali a i e impac o in luen ial obse a ions on he magni ude o he
ma ke -powe coe icien s.
Table 5
OLS ma ke -powe coe icien es ima es o I ish manu ac u ing indus ies, 1991-
1999;assuming common coe icien s ac oss manu ac u ing indus y sec o s bu excluding
in luen ial obse a ions (s anda d e o s in pa en heses)

1

2

3

4

52
R
0.00244
(0.00034)ss -0.04685
(0.00347)ss -0.02905
(0.00760)ss 0.00116
(0.00285) 0.00665
(0.00155)ss 0.27
Key:

= ma e ials;

= se ices;

= uel and powe ;

= “o he ” employmen ;

= indus ial employmen .
1 2 3 4
5
ss: deno es s a is ically signi ican a he 95% le el.
The mos ob ious di e ence be ween he es ima es in Table 5 and Table 3 is he subs an ial
educ ion in he magni ude o he ma ke powe coe icien s o “Ma e ials” (down by abou
60%), “Se ices” (down by abou 50%) and “Indus ial Employmen ” (down by abou 75%) and
he subs an ial inc ease in he size o he “Fuel and Powe ” coe icien which, unlike Table 3, is
now s a is ically signi ican . O e all he esul s a e quali a i ely simila o hose epo ed in
Table 3 and he pa ame e s, wi h he excep ion o he coe icien o “O he Employmen ”, a e
highly signi ican 8.
A second es ima ion app oach ha can handle he p oblem o in luen ial obse a ions is he
employmen o he Leas Absolu e Di e ence o LAD es ima o 9.(see Judge e al (1998)) The
es ima o minimises he impac o ou lie s by i ing he eg ession line h ough he median o he
da a a he han he mean in he case o OLS.
The LAD es ima es, inco po a ing he assump ion o cons an wo-digi sec o slopes, a e
exhibi ed in Table 6.
Table 6
7 I would like o acknowledge my colleague, Mau ice Roche’s, sugges ion o use his app oach. The p ocedu e
esul ed in abou 90 obse a ions being disca ded ou o he o al o 872.
8 We also an he model o a iable-slope coe icien s on he unca ed sample and ob ained esul s o a simila
na u e.
17
9 I would like o acknowledge my colleague, Donal O’Neill’s, sugges ion o use his es ima o .
Leas Absolu e Dis ance ma ke -powe coe icien es ima es o I ish manu ac u ing
indus ies, 1991-1999;assuming common coe icien s ac oss manu ac u ing indus y sec o s
(s anda d e o s in pa en heses)

1

2

3

4

52
R
0.00217
(0.00026)ss -0.02666
(0.00198)ss -0.00578
(0.00430) -0.00422
(0.00201)ss 0.00500
(0.00095)ss 0.32
Key:

= ma e ials;

= se ices;

= uel and powe ;

= “o he ” employmen ;

= indus ial employmen .
1 2 3 4 5
ss: deno es s a is ically signi ican a he 95% le el.
These esul s a e quali a i ely e y simila o hose epo ed in Table 5 whe e we unca ed he
sample o exclude ou lie s. Rela i e o he OLS es ima es o he comple e sample in Table 3, we
no e ha he ma ke -powe coe icien s in espec o “Ma e ials” and “Se ices” a e abou 70%
lowe in absolu e magni ude while he coe icien o “Indus ial Employmen ” is a ound 80%
lowe . The coe icien alue o “O he Employmen ” is no oo dissimila as be ween he LAD
and OLS es ima es o he comple e sample and is, in any case, also s a is ically insigni ican .
In Table 7 we p esen he es ima ed ma ke powe coe icien s gene a ed by he LAD es ima o
bu now allowing he slope coe icien s o a y by wo-digi sec o .
18
Table 7
LAD ma ke -powe coe icien es ima es o I ish manu ac u ing indus ies, 1991-
1999;allowing o di e en coe icien s ac oss he wo-digi manu ac u ing indus y sec o s
(s anda d e o s in pa en heses)
Sec o

1

2

3

4

5
15
Food 0.0016552
(0.00075)ss -0.011826
(0.01042) -0.024165
(0.02347) -0.0030469
(0.00950) 0.0084302
(0.00623)
17
Tex iles 0.0053564
(0.00127)ss -0.065583
(0.01103)ss -0.035388
(0.01359)ss 0.016506
(0.01844) 0.003349
(0.00433)
18
Wea ing
Appa el
0.014926
(0.00173)ss -0.047181
(0.01092)ss 0.068821
(0.01690)ss -0.051582
(0.01485)ss -0.0029508
(0.00421)
20
Wood &
Wood
P ods.
0.0015771
(0.00410) -0.043917
(0.01589)ss -0.0002383
(0.03255) 0.000263
(0.02368) 0.009374
(0.00983)
21
Pulp &
Pape
0.0040965
(0.00116)ss 0.0001882
(0.02268) -0.019234
(0.033691) -0.015521
(0.00966)s 0.0032912
-0.0069
22
P in ing &
Rec. Media
0.0004376
(0.00308) -0.0062446
(0.01914) 0.0020526
(0.06693) -0.004496
(0.00468) 0.0016521
(0.00322)
24
Chemicals 0.0023853
(0.00113)ss -0.033041
(0.00943)ss 0.0038951
(0.01007) -0.002727
(0.00857) -0.0044804
(0.00860)
25
Rubbe &
Plas ic
0.0014602
(0.00227) -0.0079353
(0.02279) 0.000846
(0.05596) 0.002638
(0.01800) -0.0010146
(0.01258)
26
O he Non-
Me allic
0.004832
(0.00111)ss -0.059132
(0.00411)ss -0.01114
(0.00707)s -0.011089
(0.00565)ss 0.0054426
(0.00413)
28
Fab ica ed
Me als
0.0018641
(0.00081)ss -0.032149
(0.00561)ss -0.0056407
(0.02017) 0.0035323
(0.00760) 0.0044513
(0.00455)
29
Machine y
&
Equipmen
0.0014987
(0.00111) -0.014513
(0.01291) -0.0063647
(0.03917) 0.0006608
(0.00956) 0.00000
(0.00404)
31
Elec ical
Machine y
0.0016192
(0.00109)s -0.017278
(0.01037)s 0.016068
(0.03639) 0.0010228
(0.00937) 0.000941
(0.00506)
32
Radio, TV
& Comm.
Equip.
0.0038342
(0.00092)ss -0.061462
(0.03073)ss 0.023516
(0.12890) 0.017355
(0.01102)s -0.0026887
(0.01097)
33
Med., P ec.
& Op .
Ins um.
-0.0060158
(0.00333)s -0.20872
(0.03080)ss -0.1153
(0.09407) 0.038914
(0.02731) 0.02051
(0.01683)
34
Mo o
Vehicles
-0.0011962
(0.00228) -0.023278
(0.01552)s -0.042503
(0.05921) 0.017655
(0.03136) 0.018413
(0.00472)ss
35
O he
T ans.
Equip.
0.0024885
(0.00086)ss -0.037978
(0.00463)ss 0.050381
(0.02249) -0.064264
(0.00787)ss 0.020924
(0.00284)ss
36
Fu ni u e 0.0040022
('0.00213)s 0.0005567
(0.01524) -0.024547
(0.04561) -0.0057968
(0.01696) -0.0014583
(0.00819)
19
Key:

= ma e ials;

= se ices;

= uel and powe ;

= “o he ” employmen ;

= indus ial employmen .
1 2 3 4 5
2
R
= 0.28.
s: deno es s a is ically signi ican a he 90% le el.
ss: deno es s a is ically signi ican a he 95% le el.
I we con as hese esul s o hose epo ed in Table 4 o he comple e-sample OLS es ima es,
we no e ha he LAD es ima es essen ially educe he magni ude o hose coe icien s ha would
appea o be ou lie s in he OLS se o es ima es. Thus we ind a easonably close
co espondence be ween he es ima es o “Ma e ials” wi h he excep ion o sec o s 34 (“Mo o
Vehicles ”), 35 (“O he T anspo Equipmen ”) and 36 (“Fu ni u e”). Fo “Se ices”, we
obse e ha he p edominance o nega i ely-signed coe icien s is eplica ed bu he e is a much
lowe conco dance in he wo se s o es ima es, al hough he e is a no able endency o he LAD
es ima es o be sys ema ically lowe han he OLS es ima es. The coe icien s o “Indus ial
Employmen ” a e quali a i ely e y simila o he OLS es ima es, al hough we can now only
epo s a is ically signi ican coe icien s o sec o s 34 (“Mo o Vehicles ”) and 35 (“O he
T anspo Equipmen ”). The OLS es ima es p oduce coe icien s o subs an ially g ea e
magni ude o sec o s 17 (“Tex iles”), 28 (“Fab ica ed Me als ”), 35 (“O he T anspo
Equipmen ”) and 36 (“Fu ni u e ”).
As an illus a ion o he na u e and scale o he di e ences ob ained be ween he LAD and OLS
es ima es o he comple e sample we depic he coe icien s ob ained o bo h es ima o s in he
case o “Indus ial Employmen ” in Cha 110.
10 Essen ially simila pic u es a e ob ained o he o he inpu s.
Cha 1
LAD and OLS (Comple e Sample) Ma ke Powe Coe icien s o "Indus ial
Employmen " by Two-Digi Manu ac u ing Sec o
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
15 17 18 20 21 22 24 25 26 28 29 31 32 33 34 35 36
Two-Digi Sec o
LAD
OLS
3.2.4 Robus ness checks: economies o scale
We ha e no alluded so a o he possibili y ha he indings o impe ec compe i ion could be
a bes con amina ed by, o , a wo s be mis aken o , scale e ec s. Ou eluc ance o con on
his conund um so a is because o he huge di icul y in sa is ac o ily dis inguishing empi ically
be ween impe ec compe i ion and scale e ec s wi h he Hall-Roege me hodology. Kee (2002)
shows ha i is in p inciple possible o de i e an exp ession o he di e ence be ween he Solow
quan i y and p ice-based esiduals which is a unc ion o a ma ke powe and a scale e m. Thus
by modi ying equa ion (13) as ollows
21


 

)]())[(1( 11
2
ypwxSzy kj
M
jj (16)
whe e, S is he ac ual sum o inpu sha es which may di e om uni y and k =1,…, M, i would
appea possible o simul aneously es om depa u es om compe i i e p icing and cons an
e u ns o scale. Un o una ely, equa ion (16) is non-es imable due o pe ec collinea i y be ween
he igh -hand-side a iables as will be appa en upon examina ion o equa ion (10)11.
The bes we belie e ha can be done is o add a scale e m o he speci ica ion in equa ions (14)
o (15) which will no be pe ec ly collinea wi h he ma ke -powe e m. Thus we e- an
equa ion (14) using he LAD es ima o and added he log-change in he alue o g oss ou pu
(

p+

y) as an addi ional igh -hand-side a iable. The esul s a e gi en in Table 8. I is appa en
ha he inclusion o he ou pu e m does no a ec he magni ude o he ma ke -powe
coe icien s and is i sel no s a is ically signi ican . Howe e , i has o be no ed ha his equa ion
will also su e om mul icollinea i y so we canno be empha ic abou he appa en non-
impo ance o scale e ms12.
Table 8
LAD ma ke -powe coe icien es ima es o I ish manu ac u ing indus ies, 1991-
1999;assuming common coe icien s ac oss manu ac u ing indus y sec o s and including a
scale e m (s anda d e o s in pa en heses)

1

2

3

4

5(

p+

y 2
R
0.002159
(0.00027)ss 0.027246
(0.001985)ss -0.005548
(0.004386) -0.00461
(0.002029)ss 0.005131
(0.00095)ss 0.000042
(0.00014) 0.32
Key:

= ma e ials;

= se ices;

= uel and powe ;

= “o he ” employmen ;

= indus ial employmen .
1 2 3
4 5
ss: deno es s a is ically signi ican a he 95% le el.
A less igo ous a emp o explo e po en ial scale e ec s was implemen ed by examining he
ela ionship be ween he es ima ed ma ke powe coe icien s a he wo-digi le el and indica o s
11 Kee (2002) a gues ha equa ion will end o ze o because “… he sha es o inpu in o al e enue a e mos ly
cons an in he eal wo ld …” (p. 11). As i happens his is no ue wi h ou da ase bu in any e en he e is no need
o d aw on such a s ylised ac o mo i a e why (16) canno be es ima ed gi en he a gumen made in he ex .
12 I would na u ally be p e e able o include a demand e m as a scale a iable bu none would appea o be
a ailable a he ou -digi manu ac u ing-indus y le el.
22
o scale such as he a e age numbe s employed pe local manu ac u ing uni in each sec o 13.
This analysis could es ablish no sys ema ic ela ionship. Typical o he kind o ela ionship ha
was e ealed is gi en in Cha 2 o he case o he “Ma e ials” inpu .
Cha 2
LAD Ma ke Powe Coe icien s ("Ma e ials") and Employmen by
Two-Digi Sec o
-100
-50
0
50
100
150
200
0 50 100 150 200 250 300
A g. Size o Sec o (Employmen )
Mk . Powe Coe s. x 10,000
13 We also used he le el o g oss ou pu pe local manu ac u ing uni in each sec o bu he indings we e i ually
23
4. Concluding ema ks
This pape has applied Roege ’s e sion o Hall’s ingenious non-pa ame ic es o ma ke powe
o I ish manu ac u ing ou -digi indus ies. The pape demons a es he equi alence o so-called
“monopoly” and “monopsony” e sions o he es and sugges s ha es s o “ma ke powe ” is a
mo e neu al and accep able e m. The pape also p oposes a simple es ha is capable o
disc imina ing be ween monopolis ic and monopsonis ic depa u es om compe i i e p icing.
A numbe o he indings a e wo h no ing. We ind e idence o s a is ically signi ican ma ke
powe in he p icing o key p oduc ion inpu s. We s ongly ejec he cons ancy o slopes a he
wo-digi sec o le el. Ou esul s also sugges ha he sou ce o ma ke powe is mos likely o
a ise in he p icing o inpu s a he han ou pu s. This inding con i ms he s ong p io o
p icing- aking beha iou in ou pu ma ke s which has been con i med in se e al p e ious
s udies.
We do, howe e , ind e idence ha ou es ima es o he deg ee o ma ke powe a e a ec ed o a
signi ican ex en by in luen ial obse a ions. While he s a is ical signi icance o he ma ke -
powe coe icien s is no unduly a ec ed by he employmen o es ima ion app oaches ha ake
accoun o such obse a ions, he magni udes o he coe icien s a e a ec ed mainly in a
downwa d di ec ion and especially o a hand ul o 2-digi sec o s.
While acknowledging he unsa is ac o y na u e o ou a emp o allow o scale e ec s, we
ailed o unco e any s ong e idence ha ou es ima es o ma ke powe we e con amina ed by
scale e ec s. This is an issue ha me i s u he in es iga ion.
Based on he Leas Absolu e Dis ance es ima o , ou indings s ongly sugges ha I ish
manu ac u ing indus ies, aken as a whole, exe cise s a is ically signi ican ma ke powe , albei
o a modes ex en , in pu chases o “Ma e ials” and “Indus ial Employmen ”. In he case o
“Ma e ials” we epo an o e all ma ke -powe coe icien o abou 0.2% and o “Indus ial
Employmen ” we ind a alue o abou 0.5%.
iden ical.
24
These coe icien s a e e iden ly ela i ely modes in absolu e size. Fo ins ance, i would ap
e y imp obable ha he ma ke -powe coe icien o “Indus ial Employmen ” would be
su icien in gene al o cause employmen o inc ease in esponse o inc eases in he minimum
wage. We wouldn’ be as con iden in his expec a ion o he hand ul o sec o s (“Food”, “W
and Wood P oduc s”, “Medical, P ecision and Op ical Ins umen s”, “Mo o Vehicles” and
pea
ood
O he T anspo Equipmen ”) whe e he ma ke -powe coe icien is much g ea e han 0.5%.
Rela i e o 1999 ou lays
is would sugges an annual cos bu den o indus ies o abou €30 m..
d
-
owe coe icien s a y wi hin hose wo-digi sec o s ha appea o ha e p ominen alues.
indeed
he he he Au ho i y should be unduly conce ned a all by monopsony- ype beha iou !
“
In many espec s he mos su p ising esul is he s a is ically signi ican and ela i ely la ge
nega i e ma ke -powe coe icien ha is ob ained o “Se ices”14. Fo he ull se o
manu ac u ing indus ies, he coe icien implies ha he p ice paid o his inpu is abou 2.7%
less han he in e nal ma ginal alue o his inpu o he sec o as whole.
h
Ou analysis allowed he ma ke -powe coe icien s o a y ac oss each wo-digi sec o . In
espec o he “Ma e ials” and “Indus ial Employmen ” coe icien s we ind ha a hand ul o
sec o s a e highly in luen ial in de e mining he o e all coe icien o he ull se o sec o s. A
g ea e consis ency in he coe icien alues ac oss he 17 wo-digi sec o s is, howe e , e eale
o he “Se ices” inpu . I would seem o be wo h examining he ex en o which he ma ke
p
A inal hough is whe he he o de s o magni ude o depa u e om compe i i e inpu p icing
ha a e unco e ed in his pape should a ac he in e es o he Compe i ion Au ho i y o
w
14 The coe icien o “O he Employmen ” is also nega i e bu i is s a is ically much less signi ican han
“Se ices”, “Ma e ials” o “Indus ial Employmen ”.
25