Description
The purpose of this paper is to measure the impact of HRM on organisational performance in the context of Greece. Data were collected from 178 organisations using a questionnaire survey in the Greek manufacturing sector, and analysed using the 'structural equation modelling' methodology.
Measuring the impact of HRM on
organisational performance
Anastasia A.
Katou
University of Macedonia
(GREECE)
Received September
2008 Accepted
December 2008
Abstract The purpose of this paper is to measure the impact of
HRM on organisational
performance in the context of Greece. Data were collected from 17
organisations using a !uestionnaire sur"e# in the Gree$
manufacturing sector% and anal#sed using the &structural e!uation
modelling& methodolog#. The results indicated that the relationship
'etween HRM policies (resourcing and de"elopment% compensation
and incenti"es% in"ol"ement and )o' design* and organisational
performance is partiall# mediated through HRM outcomes (s$ills%
attitudes% 'eha"iour*% and it is in+uenced '# 'usiness strategies
(cost% !ualit#% inno"ation*. Thus% the contri'ution of this stud# for
academics and practitioners is that HRM policies associated with
'usiness strategies will a,ect organisational performance through
HRM outcomes.
Keywords: HRM policies% organisational
performance% Greece
1 Introduction
Over the last ten years significant steps forward have been made in identifying the
HRM - performance relationship. However, serious gaps in our understanding still
remain with respect to the causal ordering of the variables involved in the HRM -
performance relationship (Purcell, innie, Hutchinson, Rayton, ! "wart, #$$%&
'right, (ardner, Moyniham, ! )llen, #$$*+. "pecifically, in analysing the impact of
HRM on organisational performance, each of the HRM-performance lin,age models
developed complements the others by adding constructs, variables or relationships
()lca-ar, .ernande-, ! (ardey, #$$*+. ) serious limitation that recent reviews of
Measuring the impact of HRM on organisational performance 119
A.A. Katou
the literature points out is that the lin, between HRM and organisational
performance is considered li,e a /black box/, i.e., lac, of clarity regarding /what
e0actly leads to what/ ((erhart, #$$*+. 1t was further argued that it is important to
consider the intervening steps in the HRM-performance relationship, or to consider
the variables mediating or moderating the endpoint variables (2ec,er ! (erhart,
3445+.
.urthermore, 2oselie, 6iet- and 2oon (#$$*+, by analysing the literature over the
last years on the HRM-performance relationship, reported wide disparities in the
treatment of the components emphasising the 7blac, bo07 stage between HRM and
performance. 8hey indicated that the theoretical framewor,s which dominated the
field were the 7contingent framewor,7 (i.e., HRM influences performance in relation to
contingent factors such as business strategies+ ("chuler ! 9ac,son, 34:;+, the
resource-based view (i.e., HRM influences performance according to the human and
social capital held by the organisation+ (2arney, 3443+ and the )MO theory (i.e.,
HRM influences performance in relation to employees/ )bility, Motivation and
Opportunity to participate+ ()ppelbaum, 2ailey, 2erg, ! alleberg, #$$$+.
Moreover, considering that there was no agreement on the HRM practices, policies,
and systems employed, and accordingly the constructs developed, 2oselie et al.
(#$$*+, 8obacco products
* 8e0tiles and te0tile products
5 Pharmaceuticals
3* Rubber and plastic products
35 Con-metallic mineral products
3; 2asic metal industries
3: Metal products. e0cept machinery
34 Machinery and eFuipment
#$ Office machinery and computers
#3 Glectrical eFuipment ##
Glectrical machinery
!re"uency
%3
3$
$
$
3#
#
#$
#
3
#
#
#
4
;
33
#3
%
3#
5
:
%
#
Percent
3;.>
*.5
$.$
$.$
5.;
3.3
33.#
3.3
$.5
3.3
3.3
3.3
*.3
%.4
5.#
33.:
3.;
5.;
%.>
>.*
3.;
3.3
!re"uency
*#4
;5
34
5
##4
#3
%$%
>:
*
*$
;#
3$
::
:>
35>
%$#
3:
#;$
:5
33*
;4
53
Percent
3;.4
#.5
$.5
$.#
;.:
$.;
3$.%
3.5
$.#
3.;
#.>
$.%
%.$
#.:
*.5
3$.#
$.5
4.#
#.4
%.4
#.;
#.3
#% Motopmentother transport r
and
eFui
#> .urniture
#* Other
8otal
%
5
%
3;:
3.;
%.>
3.;
3$$.$
44
3$%
33#
#4>4
%.>
%.*
%.:
3$$
8able 3. 7Main activities of the organisations in sample and population7.
Most of the Fuestions for the survey were drawn from e0isting international HRM
surveys, such as the Price 'aterhouse =ranfield ProEect "urvey (2rewster !
Hegewisch, 344>+ and Hall and 8orrington (344:+, and generally from Phillips
(3445+, Othman (3445+, 2udhwar and "parrow (344;+, "an--Halee, "abater-
"anche-, and )ragon-"anche- (3444+, and Richardson and 8hompson (3444+. 8he
Measuring the impact of HRM on organisational performance 125
A.A. Katou
Fuestionnaire was originally developed in Gnglish, then, it was translated into
(ree,, and finally translated bac, from (ree, to Gnglish. 8he translated
Fuestionnaire was piloted in 3$ organisations, and it was handed to the =GO, or
Personnel Officers, or .inancial Officers of the sample organisations due to the
reluctance of (ree, respondents to complete and return the Fuestionnaires and
due to the fact that employees are flooded with Fuestionnaires. However, the
survey Fuestionnaire was completed by one person responsible in each firm for the
HRM function. 'e ac,nowledge this as a limitation. However, the application of
Harman/s single factor test (Harman, 345;+ to all the relevant variables in the
model, using the eigenvalue greater than one criterion, revealed seven factors, and
not Eust one, with the first factor e0plaining #5.4 percent of the variance in the data
which is not relatively very high. )ccording to this test if a significant amount of
common method bias e0ists in the data, then the factor analysis of all the relevant
variables in the model will generate a single factor that accounts for most of the
variance. 8hus, we believe that the common method bias in the data was relatively
limited.
3.2 Measures
)ll variables used in the study are presented in column one of 8able #. "pecificallyD
%rganisational per&orance 'ariables
'e used multiple organisational performance variables (=henhall ! .4#:
*5.:#%
:$.:3$
8able #. 7Hariables of the survey instrument7.
"ontrols
8he control variables included in the analysis are as followsD /size' (in natural
logarithms+& /capital intensity' (natural logarithm of total assets by total
employment+&/industry' is represented by a two level variable, classifying ;:
organisations as being /traditional/ (food products, beverages, te0tiles and te0tile
products, linen, wearing apparel, footwear, and leather products+ because the
primary inputs for their production come from the agricultural sector, which is the
traditional sector in (reece, and the rest 3$$ organisations as being /non-
agricultural/ (chemical, metallic products, office machinery, electrical eFuipment,
etc.+ because the primary inputs for their production do not come from the
agricultural sector& the 'degree o& unionisation' is represented by the proportion of
staff in the organisation that is members of a trade union (i.e., 3 I $K, # I 3-
#*K, % I #5-*$K, > I *3-;*K, and * I ;5-3$$K+.
Measuring the impact of HRM on organisational performance 128
A.A. Katou
3.3 Consistency o# the sur*ey instrument
=onstruct internal consistency was chec,ed by computing =ronbach (34*3+ alphas.
8he figures in 8able # indicate that the survey instrument is a reliable instrument
for chec,ing the model presented in .igure 3, because all =ronbach alphas, e0cept
for the /business strategies/ construct, are much higher than $.;$ (Cunnally, 34;:+.
However, considering values as low as $.%* that have been found acceptable
when used with other measures (Roberts ! 'ort-el, 34;4+, we decided to maintain
the business strategy construct in spite of its low reliability. =onstruct validity was
e0amined by evaluating the percent of the total variance e0plained per dimension
obtained by applying confirmatory factor analysis (=.)+ using +. 8he percent of total variance e0plained values reported in 8able #,
are much higher than *$.$K indicating acceptable survey instrument construct
validity (Hair, )nderson, 8atham, ! 2lac,, 344*+. However, before the =.) we
preferred not carry out an e0ploratory factor analysis to identify possible structures a
priori, but instead to rely on content validity, which ensures that the measure
includes an adeFuate and representative set of items that would tap the concept
("e,aran, 344#+.
3., Statistical analysis
8o test the raised hypotheses of the proposed framewor, the methodology of
/structural e)uation odels' ("GM+ or 'latent 'ariable odels' (Hair et al., 344*+ was
used, via +. "GM is effective when testing models that are path analytic with
mediating variables, and include latent constructs that are being measured with
multiple items (, RM"G) I $.$44, C.1 I $.45, =.1 I $.4*, (.1 I
$.5:+, although fle0ible levels for RM"G) and (.1 were attained. 'e must note
Measuring the impact of HRM on organisational performance 130
A.A. Katou
here that before we conclude on the results presented in .igure # we tried all
possible paths for lin,ing controls with business strategies, HRM policies, HRM
outcomes, and organisational performance, but the results with respect to
contingencies were not significant. )s shown in .igure #D
$ffecti%eness
&'((
$fficienc
&'((
#e%elopment
&')&
Satisfaction
&')*
!nno%ation
&'(3
+ualit
&'(2
&')1 Cooperation
,it-
management
Cost
+ualit
!nno%ation
&'2&
.&'2)
.&'2(
Business
Strategies
Selection
.&')*
Organizational
Performance
&'3)
HRM Output
&'()
HRM Policies
&'/0
&')&
&')1
&')3
&'()
&')4
Cooperation
,it-
emploees
Competence
Moti%ation
Commitment
Satisfaction
Retention
1raining
2or3 "esign
Performance
appraisal
&'03
&'0*
&'04
&'0/
&'(4
Compensation
&'0(
Promotion
&'(3
!ncenti%es
&'()
&'(/
&'(&
Participation
!n%ol%ement
Communication
&'(4
&'(2
Presence
C-i.S4uare 5 1&1('32 "f 5 302 p.%alue 5 &'&&& 6orme" C-i.S4uare 5 2'0304 RMS$7 5 &'&)) C8! 5 &')* 68! 5 &')/ 98! 5 &'*(
.igure #. 78he estimated model using & Harel !
8-afrir, 3444+, it has neglected to investigate the mediating mechanisms, usually
called the 7black box7, through which HRM policies are hypothesised to affect
organisational performance. 8he testing of which HRM policies to be used in a
study in new conte0ts is of much importance (3.
)ppelbaum, G., 2ailey, 8., 2erg, P., ! alleberg, ).>%->5%.
2rewster, =., ! Hegewisch, ). (344>+. Hunan resource management in GuropeD
1ssues and opportunities. 1n =. 2rewster and ). Hegewisch (eds.+. #olicy and
practice in 1uropean huan resource anageent4 -he #rice 5aterhouse
"ran&ield 0ur'ey. #3.
alleberg, ).:-45#.
atou, ).)., ! 2udhwar, P.". (#$$5+. Human resource management systems and
organisational performanceD ) test of a mediating model in the (ree,
manufacturing conte0t. !nternational /ournal o& Huan *esource +anageent,
3;, 3##%-3#*%.
atou, ).)., ! 2udhwar, P.". (#$$;+. 8he effect of human resource management
policies on organisational performance in (ree, manufacturing firms. -hunderbird
!nternational ,usiness *e'ie., >4, 3-%*.
Measuring the impact of HRM on organisational performance 139
A.A. Katou
intana, M.5;-*$>.
'right, P.M., ! (ardner, 8. (#$$%+. 8he human resource-firm performance
relationshipD Methodological and theoretical challenges. 1n 6. Holman, 8.6. 'all,
=.'. =legg, P. "parrow, ! ). Howard (eds+. -he ne. .orkplace4 A guide to the
huan ipact o& odern .orking practices. >5.
Measuring the impact of HRM on organisational performance 141
A.A. Katou
?oundt, M., "nell, "., 6ean, 9., !
The purpose of this paper is to measure the impact of HRM on organisational performance in the context of Greece. Data were collected from 178 organisations using a questionnaire survey in the Greek manufacturing sector, and analysed using the 'structural equation modelling' methodology.
Measuring the impact of HRM on
organisational performance
Anastasia A.
Katou
University of Macedonia
(GREECE)
Received September
2008 Accepted
December 2008
Abstract The purpose of this paper is to measure the impact of
HRM on organisational
performance in the context of Greece. Data were collected from 17
organisations using a !uestionnaire sur"e# in the Gree$
manufacturing sector% and anal#sed using the &structural e!uation
modelling& methodolog#. The results indicated that the relationship
'etween HRM policies (resourcing and de"elopment% compensation
and incenti"es% in"ol"ement and )o' design* and organisational
performance is partiall# mediated through HRM outcomes (s$ills%
attitudes% 'eha"iour*% and it is in+uenced '# 'usiness strategies
(cost% !ualit#% inno"ation*. Thus% the contri'ution of this stud# for
academics and practitioners is that HRM policies associated with
'usiness strategies will a,ect organisational performance through
HRM outcomes.
Keywords: HRM policies% organisational
performance% Greece
1 Introduction
Over the last ten years significant steps forward have been made in identifying the
HRM - performance relationship. However, serious gaps in our understanding still
remain with respect to the causal ordering of the variables involved in the HRM -
performance relationship (Purcell, innie, Hutchinson, Rayton, ! "wart, #$$%&
'right, (ardner, Moyniham, ! )llen, #$$*+. "pecifically, in analysing the impact of
HRM on organisational performance, each of the HRM-performance lin,age models
developed complements the others by adding constructs, variables or relationships
()lca-ar, .ernande-, ! (ardey, #$$*+. ) serious limitation that recent reviews of
Measuring the impact of HRM on organisational performance 119
A.A. Katou
the literature points out is that the lin, between HRM and organisational
performance is considered li,e a /black box/, i.e., lac, of clarity regarding /what
e0actly leads to what/ ((erhart, #$$*+. 1t was further argued that it is important to
consider the intervening steps in the HRM-performance relationship, or to consider
the variables mediating or moderating the endpoint variables (2ec,er ! (erhart,
3445+.
.urthermore, 2oselie, 6iet- and 2oon (#$$*+, by analysing the literature over the
last years on the HRM-performance relationship, reported wide disparities in the
treatment of the components emphasising the 7blac, bo07 stage between HRM and
performance. 8hey indicated that the theoretical framewor,s which dominated the
field were the 7contingent framewor,7 (i.e., HRM influences performance in relation to
contingent factors such as business strategies+ ("chuler ! 9ac,son, 34:;+, the
resource-based view (i.e., HRM influences performance according to the human and
social capital held by the organisation+ (2arney, 3443+ and the )MO theory (i.e.,
HRM influences performance in relation to employees/ )bility, Motivation and
Opportunity to participate+ ()ppelbaum, 2ailey, 2erg, ! alleberg, #$$$+.
Moreover, considering that there was no agreement on the HRM practices, policies,
and systems employed, and accordingly the constructs developed, 2oselie et al.
(#$$*+, 8obacco products
* 8e0tiles and te0tile products
5 Pharmaceuticals
3* Rubber and plastic products
35 Con-metallic mineral products
3; 2asic metal industries
3: Metal products. e0cept machinery
34 Machinery and eFuipment
#$ Office machinery and computers
#3 Glectrical eFuipment ##
Glectrical machinery
!re"uency
%3
3$
$
$
3#
#
#$
#
3
#
#
#
4
;
33
#3
%
3#
5
:
%
#
Percent
3;.>
*.5
$.$
$.$
5.;
3.3
33.#
3.3
$.5
3.3
3.3
3.3
*.3
%.4
5.#
33.:
3.;
5.;
%.>
>.*
3.;
3.3
!re"uency
*#4
;5
34
5
##4
#3
%$%
>:
*
*$
;#
3$
::
:>
35>
%$#
3:
#;$
:5
33*
;4
53
Percent
3;.4
#.5
$.5
$.#
;.:
$.;
3$.%
3.5
$.#
3.;
#.>
$.%
%.$
#.:
*.5
3$.#
$.5
4.#
#.4
%.4
#.;
#.3
#% Motopmentother transport r
and
eFui
#> .urniture
#* Other
8otal
%
5
%
3;:
3.;
%.>
3.;
3$$.$
44
3$%
33#
#4>4
%.>
%.*
%.:
3$$
8able 3. 7Main activities of the organisations in sample and population7.
Most of the Fuestions for the survey were drawn from e0isting international HRM
surveys, such as the Price 'aterhouse =ranfield ProEect "urvey (2rewster !
Hegewisch, 344>+ and Hall and 8orrington (344:+, and generally from Phillips
(3445+, Othman (3445+, 2udhwar and "parrow (344;+, "an--Halee, "abater-
"anche-, and )ragon-"anche- (3444+, and Richardson and 8hompson (3444+. 8he
Measuring the impact of HRM on organisational performance 125
A.A. Katou
Fuestionnaire was originally developed in Gnglish, then, it was translated into
(ree,, and finally translated bac, from (ree, to Gnglish. 8he translated
Fuestionnaire was piloted in 3$ organisations, and it was handed to the =GO, or
Personnel Officers, or .inancial Officers of the sample organisations due to the
reluctance of (ree, respondents to complete and return the Fuestionnaires and
due to the fact that employees are flooded with Fuestionnaires. However, the
survey Fuestionnaire was completed by one person responsible in each firm for the
HRM function. 'e ac,nowledge this as a limitation. However, the application of
Harman/s single factor test (Harman, 345;+ to all the relevant variables in the
model, using the eigenvalue greater than one criterion, revealed seven factors, and
not Eust one, with the first factor e0plaining #5.4 percent of the variance in the data
which is not relatively very high. )ccording to this test if a significant amount of
common method bias e0ists in the data, then the factor analysis of all the relevant
variables in the model will generate a single factor that accounts for most of the
variance. 8hus, we believe that the common method bias in the data was relatively
limited.
3.2 Measures
)ll variables used in the study are presented in column one of 8able #. "pecificallyD
%rganisational per&orance 'ariables
'e used multiple organisational performance variables (=henhall ! .4#:
*5.:#%
:$.:3$
8able #. 7Hariables of the survey instrument7.
"ontrols
8he control variables included in the analysis are as followsD /size' (in natural
logarithms+& /capital intensity' (natural logarithm of total assets by total
employment+&/industry' is represented by a two level variable, classifying ;:
organisations as being /traditional/ (food products, beverages, te0tiles and te0tile
products, linen, wearing apparel, footwear, and leather products+ because the
primary inputs for their production come from the agricultural sector, which is the
traditional sector in (reece, and the rest 3$$ organisations as being /non-
agricultural/ (chemical, metallic products, office machinery, electrical eFuipment,
etc.+ because the primary inputs for their production do not come from the
agricultural sector& the 'degree o& unionisation' is represented by the proportion of
staff in the organisation that is members of a trade union (i.e., 3 I $K, # I 3-
#*K, % I #5-*$K, > I *3-;*K, and * I ;5-3$$K+.
Measuring the impact of HRM on organisational performance 128
A.A. Katou
3.3 Consistency o# the sur*ey instrument
=onstruct internal consistency was chec,ed by computing =ronbach (34*3+ alphas.
8he figures in 8able # indicate that the survey instrument is a reliable instrument
for chec,ing the model presented in .igure 3, because all =ronbach alphas, e0cept
for the /business strategies/ construct, are much higher than $.;$ (Cunnally, 34;:+.
However, considering values as low as $.%* that have been found acceptable
when used with other measures (Roberts ! 'ort-el, 34;4+, we decided to maintain
the business strategy construct in spite of its low reliability. =onstruct validity was
e0amined by evaluating the percent of the total variance e0plained per dimension
obtained by applying confirmatory factor analysis (=.)+ using +. 8he percent of total variance e0plained values reported in 8able #,
are much higher than *$.$K indicating acceptable survey instrument construct
validity (Hair, )nderson, 8atham, ! 2lac,, 344*+. However, before the =.) we
preferred not carry out an e0ploratory factor analysis to identify possible structures a
priori, but instead to rely on content validity, which ensures that the measure
includes an adeFuate and representative set of items that would tap the concept
("e,aran, 344#+.
3., Statistical analysis
8o test the raised hypotheses of the proposed framewor, the methodology of
/structural e)uation odels' ("GM+ or 'latent 'ariable odels' (Hair et al., 344*+ was
used, via +. "GM is effective when testing models that are path analytic with
mediating variables, and include latent constructs that are being measured with
multiple items (, RM"G) I $.$44, C.1 I $.45, =.1 I $.4*, (.1 I
$.5:+, although fle0ible levels for RM"G) and (.1 were attained. 'e must note
Measuring the impact of HRM on organisational performance 130
A.A. Katou
here that before we conclude on the results presented in .igure # we tried all
possible paths for lin,ing controls with business strategies, HRM policies, HRM
outcomes, and organisational performance, but the results with respect to
contingencies were not significant. )s shown in .igure #D
$ffecti%eness
&'((
$fficienc
&'((
#e%elopment
&')&
Satisfaction
&')*
!nno%ation
&'(3
+ualit
&'(2
&')1 Cooperation
,it-
management
Cost
+ualit
!nno%ation
&'2&
.&'2)
.&'2(
Business
Strategies
Selection
.&')*
Organizational
Performance
&'3)
HRM Output
&'()
HRM Policies
&'/0
&')&
&')1
&')3
&'()
&')4
Cooperation
,it-
emploees
Competence
Moti%ation
Commitment
Satisfaction
Retention
1raining
2or3 "esign
Performance
appraisal
&'03
&'0*
&'04
&'0/
&'(4
Compensation
&'0(
Promotion
&'(3
!ncenti%es
&'()
&'(/
&'(&
Participation
!n%ol%ement
Communication
&'(4
&'(2
Presence
C-i.S4uare 5 1&1('32 "f 5 302 p.%alue 5 &'&&& 6orme" C-i.S4uare 5 2'0304 RMS$7 5 &'&)) C8! 5 &')* 68! 5 &')/ 98! 5 &'*(
.igure #. 78he estimated model using & Harel !
8-afrir, 3444+, it has neglected to investigate the mediating mechanisms, usually
called the 7black box7, through which HRM policies are hypothesised to affect
organisational performance. 8he testing of which HRM policies to be used in a
study in new conte0ts is of much importance (3.
)ppelbaum, G., 2ailey, 8., 2erg, P., ! alleberg, ).>%->5%.
2rewster, =., ! Hegewisch, ). (344>+. Hunan resource management in GuropeD
1ssues and opportunities. 1n =. 2rewster and ). Hegewisch (eds.+. #olicy and
practice in 1uropean huan resource anageent4 -he #rice 5aterhouse
"ran&ield 0ur'ey. #3.
alleberg, ).:-45#.
atou, ).)., ! 2udhwar, P.". (#$$5+. Human resource management systems and
organisational performanceD ) test of a mediating model in the (ree,
manufacturing conte0t. !nternational /ournal o& Huan *esource +anageent,
3;, 3##%-3#*%.
atou, ).)., ! 2udhwar, P.". (#$$;+. 8he effect of human resource management
policies on organisational performance in (ree, manufacturing firms. -hunderbird
!nternational ,usiness *e'ie., >4, 3-%*.
Measuring the impact of HRM on organisational performance 139
A.A. Katou
intana, M.5;-*$>.
'right, P.M., ! (ardner, 8. (#$$%+. 8he human resource-firm performance
relationshipD Methodological and theoretical challenges. 1n 6. Holman, 8.6. 'all,
=.'. =legg, P. "parrow, ! ). Howard (eds+. -he ne. .orkplace4 A guide to the
huan ipact o& odern .orking practices. >5.
Measuring the impact of HRM on organisational performance 141
A.A. Katou
?oundt, M., "nell, "., 6ean, 9., !