Upload
independent
View
0
Download
0
Embed Size (px)
Citation preview
Training and organisational innovations in a
local industrial system: empirical evidence
from Emilia-Romagna1
Giovanni Guidetti, Department of Economics, University of BolognaMassimiliano Mazzanti, Department of Economics, Institutions andTerritory, University of FerraraHuman Resource Management Journal, Vol 17, no 3, 2007, pages 283–306
The article studies the driving forces of firm training using a survey-based datasetof manufacturing firms in the Emilia-Romagna region, Northern Italy. The data arederived from the responses to a structured questionnaire administered in 2002 to themanagement of a representative sample of firms with more than 50 employees in thehighly industrialised province of Reggio Emilia. Firms’ training choices are analysedusing a theoretical/conceptual framework based on the notion of complementarityamong productive factors. Training is provided as long as it favours theestablishment of complementary relationships among the skills it develops and otherinputs. The main factors associated with training include structural characteristics,HRM practices, workforce features, labour management and performance of thefirm. Training activities emerge as being positively associated with organisationalpractices that affect the whole firm: workforce skill level, firm size, firm productivityand labour flexibility. The role of HRM practices in driving training is brought intoquestion. These are key issues for the current debate on the development of localsystems in the European and Italian context. The high and joint relevance ofstructural variables and labour demand-related factors shows that regionalindustrial policies must support labour policies within an integrated policy effortaimed at increasing potential firm productivity.Contact: Giovanni Guidetti, Dipartimento di Scienze Economiche, StradaMaggiore 45, 40125 Bologna, Italy. Email: [email protected]
INTRODUCTION: THE HUMAN CAPITAL APPROACH
In his seminal contribution on training in firms, Becker (1975) draws the crucialdistinction between specific and general training and analyses its consequences.Assuming perfect competition in both the labour and product markets and assumingperfect information and perfect mobility of productive factors, Becker (1975) showsthat the funding of training for the acquisition of skills/knowledge that positivelyaffects employees’ productivity in the firm financing the training, as well as in othercomparable firms, does not occur; that is, employers do not fund general training.Employers provide funding for specific training, that is, the acquisition ofknowledge/skills that will positively influence productivity in their firms. In thistype of training, the financial burden is shared by the employees, who benefit fromthis training support; they share with the employer the financial burden of the direct
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 283
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4
2DQ, UK and 350 Main St, Malden, MA, 02148, USA.
training and opportunity costs. Thus, employers and employees are both regardedas rational agents, maximising an objective function, within a set of constraints.Training activities increase employees’ productivity, and the target for bothemployees and employers is to maximise the remuneration arising from theirrent-seeking activities.
The economic literature has analysed the consequences of relaxing Becker’s (1975)strict hypotheses related to perfect information and perfect competition in both thelabour and product markets. Stevens (1994, 1999) and Acemoglu and Pischke (1999)show that the presence of turnover costs, imperfect information and imperfectcompetition in either the labour or the product market favours the profitability toemployers of financing general training. Other contributions have pursued a slightlydifferent strategy (Acemoglu and Pischke, 1999; Lazear, 2003). These analyses regardgeneral training as being specific training or being complementary to specifictraining. Both approaches see the financing of general training as being beneficial toemployers.
Economists have carried out much empirical analysis of the human capitalapproach; a detailed survey of this literature is beyond the scope of this article.However, we should mention that several articles deal with the propensity ofemployers to provide training but focus on the distinction between formal andinformal training, neglecting its degree of specificity. This is because of pooravailability of appropriate data and the difficulties involved in measuringempirically the degree of firm specificity of training programmes. In addition, almostall the empirical literature on human capital considers the firm’s structural features,such as size, sector, composition of the workforce, etc., as determinants of thepropensity to introduce training programmes. The relevance of these variables stemsfrom casual empiricism and is not explicitly rooted in any of the theoretical analysesreferred to in the previous discussion. In fact, the theoretical human capital literaturespecifically addresses the effects of deviations from the standard assumptions ofperfect competition on the behaviour of maximising agents, ignoring the influence ofstructural variables. The theoretical section of this article aims to fill this gap in theliterature by bridging the theoretical and empirical analyses.
The article is structured as follows. The first section develops a critical conceptualframework where training and skills are conceived as critical elements in the analysisof production processes. The second section presents the empirical evidence, basedon recent survey data, related to critical factors correlated with firm trainingstrategies and organisational practices in a local industrial system. The third sectionconcludes by examining the main implications of empirical results and providingsuggestions for regional policy.
HUMAN CAPITAL, TRAINING AND PRODUCTION: A CONCEPTUAL
FRAMEWORK
Complementarities in production
Milgrom and Roberts (1990, 1995) developed a formal model that refinesEdgeworth’s approach to complementarity among productive factors. In theircontributions, specific units of analysis are not defined; they refer either tocharacteristic features of production (Milgrom and Roberts, 1995) or to ‘elements of
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007284
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
the firm’s strategy’ (Milgrom and Roberts, 1990: 513), or more broadly, to ‘groups ofactivities (Milgrom and Roberts, 1990: 514). From a labour economics’ perspective,complementarities among productive factors can be discussed with reference to fourunits of analysis:
1. employees’ individual skills and training practices adopted for skills development;2. division, shop floor, teams or, more generically, autonomous subunits of the
productive unit;3. organisational practices related to organisation of work in a broad sense (i.e.
teamwork, task rotation, training practices) and to other defining features ofproduction (i.e. management of inventories, degree of vertical integration);
4. capital equipment such as hardware (i.e. lathes, computers) and software (i.e.computer-aided design, word-processing programs).
These four units of analysis are the inputs of production, which is conceived as aprocess of coordination of continuous and ever-changing interactions among inputs.
Complementarity and skills Complementarity among inputs means that thereturns from a single skill do not depend on that particular skill, but are related toother skills and inputs. For this reason, it is useful to introduce the distinctionbetween skills acquired and skills used. The former refers to the content ofeducation, training and, more generally, to the knowledge that is transmitted to theemployee. Skills acquired includes the employee’s stock of knowledge and previousworking experience, which are definable regardless of the specific productive contextin which she operates. Acquisition of skills occurs through formal (formal education,training) and informal transmission of knowledge. The latter applies to the skillsactually used by employees in their working activities and defines the set of tasksthey are required to perform. These types of skills cannot be specified outside awell-defined productive context, and their development can be achieved throughtypes of formal and informal training. Skills used are assets whose specificitydepends on the complementarity with other inputs.
Employees’ learning can be understood as a dynamic process of specification ofcomplementarity between the skills acquired and other inputs, which gives rise tothe set of skills used. The establishment of complementary relationships among skillsacquired and other inputs and skills that exist in the firm translates in the set of skillsused. Learning processes, such as those implied by on-the-job training, learning bydoing and other diverse training practices, specify this web of relationships amongskills acquired and other inputs. In a sense, training can be seen as a tool facilitatingthe implementation of complementary relationships among inputs.
The relationship between skills used, skills acquired and training implies that theeffect of training on individual productivity can be rather complex. As far as generaltraining is concerned, it has a direct effect on the endowment of individualknowledge and the range of skills acquired. The setting up of new complementaryrelationships specifies the effect of general training on the skills used. Therefore, itis not the content of general training, but rather the setting up of newcomplementary relationships that different forms of training can favour, whichdetermines the effect of general training on individual productivity.
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 285
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
This view of learning, and the dichotomy between skills acquired and skills used,cause the collapse of the identification between training and skills implied inBecker’s (1975) work. Ever since Becker’s (1975) seminal analysis, the distinctionbetween general and specific training has overlapped with that between general andspecific skills. Thus, specific training gives rise to firm-specific skills and generaltraining develops general skills. However, general training implies the acquisition ofgeneral skills. Skills used determines the actual range of an employee’s tasks, dutiesand productivity; the widening of an employee’s endowment of skills acquired doesnot necessarily entail an increase in the level of her productivity. Productivity isdictated by the complementary relationships set up in the firm employing theworker at the time she receives the training. Therefore, labour productivity is alwaysfirm specific because the return from the skills acquired is dependent on the highlyidiosyncratic skills used. Hence, even when the skills acquired are general, theirreturn is always firm specific.
This analysis of training, learning and skill development raises two crucialconsequences. First, general training affects productivity in the firm in which theemployee is currently employed (internal productivity), and productivity asperceived by employers in the external labour market (external productivity) indifferent ways. Divergence between internal and external productivity favours thesetting up of internal labour markets, as they insulate the employers financingtraining from underbidding by other employers. Second, the focus of analysis shiftsfrom a distinction between general and specific training to analysis ofcomplementary relationships among inputs. If general training can develop specificassets, this occurs through the interaction of this kind of training with other inputs.General training practices fit with other inputs and training practices; theirinteractions favour the process of skill development previously described. Thismeans, as far as training practices are concerned, that general training needs to beanalysed jointly with other inputs, as in Milgrom and Roberts’s (1995) analysis, inorder to understand its impact on firm productivity. It is useful to emphasise that theeffect of general training is not limited to individual productivity but has a widereffect because of complementarity with other productive inputs. Of course, this doesnot mean that employers are always willing to finance general training. However, thedistinction between skills acquired and used provides a rationale for understandingthe potential profitability of general training for employees.
Some hints about the interaction between structural variables and skill
development
To conclude this theoretical analysis, it is important to discuss the interactionbetween some firm structural variables and the process of skills developmentpreviously discussed. We focus on three different elements of the firm’s structure: (1)firm size; (2) firm technology; and (3) organisational choices and internal labourmarket.
1. Firm size. It is reasonable to believe that fewer complementary relationshipsamong inputs can be coordinated in small firms than in large firms. This wouldimply a negative impact on the productivity of small firms; if fewercomplementary relationships can be established, then the same set of acquired
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007286
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
skills will produce a lower level of returns from the skills used for small firms.This reduces the value to the employer in a small firm of providing training foremployees.
2. Firm technology. Technology constrains the conversion of skills acquired into skillsused. Capital equipment, machinery and productive processes in generalcharacterise technology. These factors constitute the productive inputs with whichthe skills acquired must establish a complementary relationship. The processof conversion of skills depends on how the match between the elementscharacterising technology and the development of skills used is coordinated andmanaged. Of course, the relationship between technology and skills works in theopposite direction, i.e. from skills used to technological development.
3. Organisational choices and internal labour market. The strategies adopted bymanagement in relation to production and work organisation affect the internalstructuring of the firm and the potential for the implementation of complementaryrelations between training practices and organisational options. Although choicesrelated to the firm’s organisation are aimed at increasing productivity levels, thisdoes not imply that development of skills is always required to implement thesechoices. Good organisational strategies can achieve higher levels of productivitywithout any further development of employees’ skills by better exploitingcomplementary relationships among existing skills or by extracting higher levelsof effort through reducing opportunistic behaviour. Thus, practices such as taskand job rotation, teamwork, quality circles (QC), etc. can give rise to an increasein the potential of complementary relationships without any enlargement of therange of employees’ skills.
Likewise, in the organisational structuring of the firm, internal labour markets do notalways provide the most suitable environment for the process of skills development,even for long-term employees. In fact, the role of tenure is ambiguous. Certainly,provision of training occurs based on expected tenure being sufficiently long for thecompany to reap the benefits of the training costs. But tenure is a necessary but nota sufficient condition for the provision of training. Employees’ trainability (Thurow,1975) and quality of labour demand play a pivotal role. If poor employee trainabilityraises training costs, or if the firm’s potential to establish profitable complementaryrelationships between newly developed skills and other inputs is scarce, then tenurealone will not induce the provision of training. Therefore, an analysis of theassociation between training and tenure will test the quality of labour supply ordemand.
Bearing in mind these considerations about the internal organisation of the firm,we can highlight some relationships between labour flexibility and provision oftraining.2 Labour flexibility is a complex notion; there is no clear-cut relationshipbetween labour flexibility and the propensity to train. As far as numerical flexibilityis concerned, the link is ambiguous. On the one hand, it might be expected that themore intensive the use of practices aimed at strengthening numerical flexibility, thelower the propensity to train employees, as numerical flexibility decreases anemployer’s possibility to profit from the increase in productivity. On the other hand,the learning mechanisms outlined in the previous section indicate that a minimumamount of training is always required to establish the complementary relationships
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 287
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
necessary for any productive activity. Therefore, any flexible staffing practices willresult in an increased requirement for training. As far as functional flexibility isconcerned, the link with training activities is also ambiguous. On the one side, onemight expect that the pursuit of functional flexibility would involve a high level oftraining of employees; on the other side, functional flexibility can result in poorspecialisation, especially in menial jobs in highly regulated labour markets. Hence,functional flexibility can substitute for training when there are restrictions onnumerical flexibility.
EMPIRICAL ANALYSIS
Context and dataset
The empirical evidence on training derives mainly from micro-based contributions,which take the worker as the unit of analysis. While there is a rich array of data ontraining from cross-sectional and longitudinal individual-based surveys, data on thenature of training investments and training typologies by establishments and firmsare scarce (Frazis et al., 1995) despite the fact that they may provide critical insightson the management of high-performance practices (HPP) in local economic systems,taking a firm-based perspective. We do not discuss the more recent papers on firmtraining (among others, Black et al., 1999; Whitfield, 2000; Beckmann, 2002), most ofwhich provide detailed information, but for only a limited set of explanatoryvariables. For other variables, information is deficient or absent, thus introducing thewell-known problem of critical factor omission. The empirical value of our study liesin its investigation of a full set of training indexes and the introduction of acomprehensive set of possibly correlated factors. Our datasets, which are presentedin succeeding discussions, allow a detailed and robust analysis of the mostsignificant variables associated with firm training. We have good measures of anumber of establishment control variables and access to good-quality data on pastperformance from official accounting statistics.
A particularly rich set of key explanatory variables and simple controls thatshould mitigate any selectivity bias is an important asset for the estimation stage(Boheim and Booth, 2004). The analysis investigates the role of training as ahigh-performance practice in critical industrial environments, exploring the nature offirm strategies by drawing out what linkages exist between training and: (1) otherHRM practices, (2) firm general characteristics, (3) labour management, (4) pastperformances and (5) techno-organisational innovations.
The analysis presented here is based on a study of the Reggio Emilia Provincein the Emilia-Romagna region. Emilia-Romagna is an area of Northern Italycharacterised by a high density of industrial districts and is home to 7 per centof the Italian population. Geographically speaking, the region links the north andcentre of Italy. The cities that are best known and most important economicallyare Bologna, Parma, Modena and Reggio Emilia, where in terms of value addedand employment, manufacturing activities are still relevant. Rimini and Ravennaare the biggest cities on the Adriatic coast and are more oriented to services. Interms of per capita GDP, the region ranks 10th among the 122 regions of theEuropean Community (EC) (based on an EC regional average GDP index of 100,the regional value is 127.6) and has an average unemployment rate of 5 per cent.
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007288
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
Half of the region’s enterprises are involved in the service sector, the remainderbeing evenly split between agriculture and manufacturing. Manufacturing is stillthe main source of economic development and includes the following sectors:chemicals, textiles, ceramics, motorcycles, packaging machinery, farmingmachinery, biomedical, wood processing, machine tools and food. Italianproduction systems are based on small- and medium-sized enterprises. The localindustrial system of Reggio Emilia is characterised by dynamism and highinnovation intensity. The present case study aims at providing new empiricalevidence on training practices and other HRM activities in local productionsystems from the viewpoint of the firm. As labour and industrial policies areimplemented at regional level, the study will inform policy makers about the needfor policies aimed at fostering economic development and targeting techno-organisational innovation practices in firms.
The sample consists of 166 firms drawn from a total of 257 companies located inthe province – listed in both national (Intermediate Census 1996 of the NationalInstitute of Statistics) and local (Camera di Commercio in Reggio Emilia 2001)databases. These firms operate in 19 manufacturing sector codes and have at least 50employees. Although 199 firms responded to the survey (the questionnaire had aresponse rate of 77.4 per cent), economic performance indicators were available foronly 166 of these firms. Economic performance indicators cover the period 1998–2001and are derived from firm balance sheets registered in the Reggio Emilia Chamberof Commerce. Our sample therefore represents 65 per cent of the population. Thedistribution of firms by sector and size is characterised by limited bias whencomparing the 166 firms with all surveyed firms (Tables A.2 and A.3 in theAppendix). The textile sector and small-sized firms (50–99 employees) are bothslightly under-represented. However, there is no significant distortion in othersectors or in other employee classes, with the number of interviewed firmsapproaching or reaching 100 per cent of the total in many cases.3
Methodology
We use different proxies for training as dependent variables: indexes of totalcoverage and indexes of training activity adoption. Dummies for training ofemployees and newly hired workers are included in the bivariate discrete analysis.
From Table 1 it can be seen that training is provided in 80 per cent of firms andthat training for newly hired employees is provided in 78 per cent of firms. Also, wecan see that 45 per cent of employees are involved in training activities. The numberof formal and informal training practices adopted is high compared to the Italianaverage: our synthetic index for the number of practices adopted presents a meanvalue of 0.61. The potential driving forces of the types of training in our sampleinclude firms’ structural characteristics, labour demand management, HRMpractices, workforce features and firm performance. The full set of dependant andexplanatory variables used is presented in Tables 1 and 2. The availability of anextended dataset that includes firm characteristics allows us to control for manyfactors that are important in decisions over training and reduces the possibledistortions that can arise in cross-sectional analyses from omitting relevantexplanatory factors.
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 289
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
TAB
LE
1Tr
aini
ngde
pend
ent
vari
able
s
Var
iab
leA
cron
ymTy
pe/
ran
geD
escr
ipti
onM
ean
valu
e
Trai
ning
for
empl
oyee
sT
RA
IN-E
MP
Dum
my
Bin
ary
vari
able
taki
ngva
lue
1if
the
firm
was
offe
ring
(in
2001
)fo
rmal
and
/or
info
rmal
trai
ning
prog
ram
mes
for
empl
oyee
s
0.80
Trai
ning
for
new
hire
dem
ploy
ees
TR
AIN
-NE
WD
umm
yB
inar
yva
riab
leta
king
valu
e1
ifth
efir
mw
asof
feri
ng(i
n20
01)
form
alan
d/
orin
form
altr
aini
ngpr
ogra
mm
esfo
rne
wly
hire
dem
ploy
ees
0.78
Trai
ning
Cov
erag
eT
RA
IN-C
OV
Con
tinu
ous
01
Perc
enta
geof
wor
kers
invo
lved
intr
aini
ng0.
45
Ind
exof
Trai
ning
typo
logi
esad
opti
onT
RA
IN-A
DO
PC
onti
nuou
s0
1T
his
inte
nsit
yin
dex
capt
ures
the
num
ber/
vari
ety
offo
rmal
and
info
rmal
trai
ning
acti
viti
esad
opte
dby
firm
s(a
dd
ing
and
aver
agin
gfiv
ety
pes:
on-t
he-jo
btr
aini
ng,
spec
ific
trai
ning
(int
erna
lco
urse
s),
gene
ral
trai
ning
(int
erna
lco
urse
s),
spec
ific
trai
ning
(ext
erna
lco
urse
s),
gene
ral
trai
ning
(ext
erna
lco
urse
s)
0.71
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007290
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
TAB
LE
2T
hese
tof
expl
anat
ory
vari
able
s
Var
iab
les
Typ
e/ra
nge
Acr
onym
Des
crip
tion
Mea
nva
lue
AFi
rmst
ruct
ural
vari
able
sA
.1Fi
rmsi
zeD
umm
ies
ME
DIU
M,
ME
DIU
M-
LA
RG
E;
LA
RG
E(s
mal
lbe
nchm
ark)
Dum
mie
sco
ncer
nfir
ms
wit
h10
0–24
9em
ploy
ees
(med
ium
);25
0–49
9(m
ediu
m-l
arge
);>5
00(l
arge
)0.
29;
0.17
;0.
12
A.2
Prod
ucti
vese
ctor
sà
laPa
vitt
Dum
mie
sL
I,R
I,SS
(sca
le-
inte
nsiv
ebe
nchm
ark)
Lab
our
inte
nsiv
e,re
sour
cein
tens
ive,
spec
ialis
edsu
pplie
rs,
scal
ein
tens
ive
0.16
;0.
28;
0.41
A.3
Firm
typo
logy
Dum
mie
sPR
IV,
CO
OP
(gro
ups
and
cons
orti
umbe
nchm
ark)
Priv
ate
firm
,co
oper
ativ
efi
rms/
coop
erat
ive
grou
p0.
63:
0.06
A.4
Shar
eof
reve
nue
ond
omes
tic
mar
kets
Con
tinu
ous
01
NA
T-R
EV
Shar
eof
reve
nue
ond
omes
tic
mar
kets
0.56
6
A.5
Shar
eof
reve
nue
from
subc
ontr
acti
ngC
onti
nuou
s0
1M
KT-
RE
VSh
are
ofre
venu
efr
omsu
bcon
trac
ting
0.19
25
BW
orkf
orce
feat
ures
B.1
Em
ploy
ees
educ
atio
n/sk
illle
vel
Con
tinu
ous
01
ED
UC
Thi
ssh
are
ind
exca
ptur
esth
eed
ucat
iona
lle
vel
cont
ent
ofth
ew
orkf
orce
.Fir
mem
ploy
men
tis
div
ided
into
blue
colla
r,w
hite
colla
r(s
kille
dan
dun
skill
ed),
and
man
ager
ial
posi
tion
s.(O
nly
theo
reti
cally
itas
sum
es0,
1va
lues
.)
0.32
B.2
Shar
eof
man
ual
wor
kers
Con
tinu
ous
01
MA
NU
AL
Shar
eof
wor
kers
empl
oyed
inm
anua
ljo
bs0.
65
CFl
exib
ility
inpr
oduc
tion
proc
ess
and
labo
urse
rvic
esC
.1Pl
ant
flex
ibili
tyC
onti
nuou
s0
1PL
AN
T-FL
EX
Ind
exca
ptur
ing
the
flex
ibili
tyin
utili
sati
onof
plan
tte
chno
logi
es.I
tad
ds
upan
dav
erag
esth
epr
esen
ceof
flex
ibili
tyel
emen
tsco
ncer
ning
the
plan
tpr
oduc
tion
proc
esse
s(i
.e.p
rod
ucti
onor
gani
sed
insh
ifts
,sha
reof
wor
kers
invo
lved
insh
ifts
,pro
duc
tion
ofla
rge
volu
mes
/sm
all
seri
eson
ly(l
ow‘d
iffer
enti
atio
n’)
or,
alte
rnat
ivel
y,bo
thki
nds
ofpr
oduc
tion
(hig
h‘d
iffer
enti
atio
n’),
pres
ence
offl
exib
lem
ulti
func
tion
plan
tsy
tem
s
0.36
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 291
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
TAB
LE
2C
onti
nued
Var
iab
les
Typ
e/ra
nge
Acr
onym
Des
crip
tion
Mea
nva
lue
C.2
Synt
heti
cin
dex
ofla
bour
flex
ibili
tyC
onti
nuou
s0
1L
AB
-FL
EX
Ad
dit
ive
ind
exth
atin
clud
esin
form
atio
non
the
use
ofsh
ort-
term
cont
ract
s,fu
ncti
onal
flex
ibili
ty,
wag
efl
exib
ility
and
inno
vati
on/
flex
ibili
tyin
wor
king
hour
regi
mes
.
0.30
C.3
Firm
hier
arch
ical
stru
ctur
e(h
iera
rchi
cal
leve
ls/
firm
func
tion
s)
Con
tinu
ous
01
HY
ER
AR
CH
iera
rchi
cal
inte
nsit
yst
ruct
ure
isd
efine
das
the
rati
oof
the
num
ber
hier
arch
ical
laye
rson
the
num
ber
offo
rmal
ised
firm
div
isio
ns.
The
ques
tion
nair
eid
enti
fied
15d
isti
nct
form
alis
edd
ivis
ions
.T
heav
erag
enu
mbe
rof
div
isio
nsis
10.5
.
0.29
DIn
dust
rial
rela
tion
sD
.1Sy
nthe
tic
ind
exof
wor
ker’
sin
volv
emen
tin
firm
man
agem
ent
init
iati
ves
Con
tinu
ous
01
INV
OLV
Com
posi
tein
dex
:in
clud
esal
lin
form
atio
nre
gard
ing
the
exte
ntto
whi
chw
orke
rsar
ein
volv
edin
prod
ucti
onan
din
nova
tion
orie
nted
dec
isio
ns:
high
erva
lues
are
asso
ciat
edw
ith
cons
ulta
tion
and
barg
aini
ngpr
oces
ses
onfi
rmd
ecis
ions
0.28
EP
erfo
rman
ceva
riab
les
(mea
nva
lues
are
used
,fo
rpe
riod
s19
95–2
001,
or19
95–1
997)
E.1
Net
profi
t/re
venu
eC
onti
nuou
sPR
OF
Bal
ance
shee
tsd
ata
(mea
nva
lues
peri
od19
95–2
001)
E.2
Val
uead
ded
per
empl
oyee
(pro
duc
tivi
ty)
Con
tinu
ous
PRO
DU
CB
alan
cesh
eets
dat
a(m
ean
valu
espe
riod
1995
–200
1)
FTe
chno
-org
anis
atio
nal
inno
vati
ons
F.1a
Synt
heti
cin
dex
ofor
gani
sati
onal
inno
vati
on(fi
vehi
gh-p
erfo
rman
cepr
acti
ces)
Con
tinu
ous
01
INN
O-O
RG
Ad
dit
ive
ind
exca
ptu
rin
gth
ein
ten
sity
ofin
nov
atio
nad
opti
ons
inte
rms
ofth
efi
veor
gani
sati
onal
prac
tice
s(F
.1b)
0.27
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007292
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
F.1b
Hig
h-pe
rfor
man
cepr
acti
ces/
orga
nisa
tion
alin
nova
tion
(qua
lity
circ
les,
team
-wor
king
,ju
st-i
n-ti
me,
task
rota
tion
,To
tal
Qua
lity
Man
agem
ent)
five
dum
mie
sQ
C,
TE
AM
,JI
T,TA
SK,
TQ
MIt
take
sth
eva
lue
1w
hen
the
firm
has
impl
emen
ted
such
orga
nisa
tion
alin
nova
tion
prac
tice
over
1998
–200
10.
12;
0.30
;0.
13;
0.32
;0.
46
F.2
Prod
uct
inno
vati
ond
umm
yIN
NO
-PR
OD
Itta
kes
the
valu
e1
whe
nth
efi
rmha
sin
trod
uced
ate
chno
logi
cal
inno
vati
onov
er19
98–2
001
0.66
F.3
Proc
ess
inno
vati
ond
umm
yIN
NO
-PR
OC
Itta
kes
the
valu
e1
whe
nth
efi
rmha
sin
trod
uced
such
tech
nolo
gica
lin
nova
tion
prac
tice
sov
er19
98–2
001
0.68
F.4
Qua
lity
prod
uct
inno
vati
ond
umm
yIN
NO
-QU
AL
Itta
kes
the
valu
e1
whe
nth
efi
rmha
sin
trod
uced
such
tech
nolo
gica
lin
nova
tion
prac
tice
sov
er19
98–2
001
0.67
F.5
Tech
nolo
gica
lIn
nova
tion
ind
ex(s
ynth
etic
ind
exof
F.2-
F.4)
Con
tinu
ous
01
INN
O-T
EC
HIn
dex
cap
turi
ng
the
inte
nsi
tyof
inn
ovat
ion
adop
tion
s(i
nte
rms
ofF.
2-F.
4d
umm
ies)
0.62
F.6
Em
ploy
eeFo
rmal
Eva
luat
ion
Con
tinu
ous
01
FOR
M-E
VA
LSh
are
ofem
ploy
ees
subj
ect
tofo
rmal
eval
uati
onpr
ogra
mm
es,
incr
easi
ngly
wei
ghte
dby
clas
ses
(fro
mlo
wsk
illed
toto
pm
anag
ers)
0.33
Not
e:C
onti
nuou
s0
1m
eans
that
the
ind
ex/
vari
able
take
sva
lues
wit
hin
the
0–1
rang
e.In
tens
ity
ind
exis
cons
truc
ted
byad
din
gan
dav
erag
ing
dic
hoto
mou
san
swer
sre
late
dto
the
pres
ence
/ab
senc
eof
inno
vati
onin
the
real
ms
ofte
chno
logy
,and
labo
ur/
prod
ucti
onor
gani
sati
on.A
llre
gres
sors
cons
ider
edar
esh
own.
Som
em
ayno
tap
pear
infi
nal
regr
essi
ons
ifth
eytu
rnou
tne
ver
sign
ifica
nt.
TQ
M=
Tota
lQ
ualit
yM
anag
emen
t;JI
T=
just
-in-
tim
e.
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 293
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
The aim of the investigation is to assess the relationship between training in firmsand its driving forces using different synthetic indexes of the main forms of trainingactivities as dependent variables.
The estimated regression is a reduced form:
Training index Firm characteristics HRM Innovatii i i i= + [ ] +β β β0 1 2 oon practicesWorkforce features Performances
[ ]+ [ ] + [ ] +β β ε3 4i i i
It should be noted that the data are cross-sectional;4 thus, the causality links betweenvariables are generally ‘weak’. The objective is not to test cause–effect relationships,but to assess the significance and intensity of relationships between the variables(Michie and Sheehan, 2005). We are able to exploit ‘lagged’ information concerningfirm performance (data are available for years 1995–2001), and data on HRM/organisational innovation (trend data for 1998–2001),5 as potential drivers of trainingcalculated on 2001 related data. Panel data, although in some ways more appropriateas unobserved heterogeneity is dropped, may be associated with problems whenfocusing on innovation and the intangible factors of production, often quasi-fixed orslowly evolving factors. However, survey-based approaches are often the only optionavailable when the aim is to elicit information from a rich and detailed datasetinvolving intangible assets, innovative dynamics and HRM practices, includingtraining, at the micro level.
The analysis necessarily relies on simple reduced forms, specifying propereconometric models for each continuous/discrete variable under analysis.6 This isthe practice usually adopted in the empirical literature on training drivers. The‘pillars’ that give the study robustness are sample representativeness, the quality andquantity of firm-level data, and the way we cope with endogeneity, omitted variableissues and other potential flaws affecting the analysis.
A preliminary analysis was undertaken to study the full correlation matrix,including all potential covariates, and dropping highly correlated potentialregressors. This first selection aimed at reducing collinearity problems and selectinga limited set of covariates for testing each specific hypothesis. Such problems areintrinsic to applied investigations exploiting data on techno-organisationalinnovation and high-performance practices, which are often highly interrelated. Theoutcome is a matrix of selected potential explanatory variables (correlation values forselected regressors are shown in Table A.1 in the Appendix).7 In addition to a fewvariable indexes, which are sequentially and cautiously introduced, the finalcorrelation matrix shows low figures for the main independent variables. Thevariables presented in Table 2, and which are included in the correlation matrix, arethose that we selected from this preliminary analysis. They are the variables weexploit as covariates: they satisfy the criterion of being more significant in terms ofproxies and are less affected by correlation problems overall.
It should also be noted that we adopted a ‘from general to particular’ backwardstepwise method for our analysis in order to tackle the possible biases arising fromomission of relevant variables or from inclusion of irrelevant ones (in the formercase, coefficients are biased; in the latter, variances are inflated as a result of too muchinformation and estimates are less efficient). We argue that the second problem, thatis, overfitting specifications starting from a conceptual model including the full
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007294
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
vector of potential training drivers, is less severe and can be resolved by eventuallydeleting non-significant variables (i.e. t ratios of less than 1.282, step by step).
Hypotheses
The theoretical framework developed in previous discussions points to a set ofpropositions concerning some critical focal points, each affecting the propensity offirms to train employees. These focal points refer to organisational traits related tothe firm’s management of human resources, to structural features, or to firmeconomic performance. For each point, the potential complementarity with trainingactivities will be discussed. The purpose of this section is to specify, for each focalpoint, those variables that can potentially establish complementary relationshipswith skill development. When possible, the expected relation will be stated; however,this is not always feasible because, especially in relation to HRM, the relation withtraining, a priori, is not always one way, but can depend on how these practices areimplemented in a specific firm. Therefore, the results of our empirical analysisshould demonstrate whether or not complementary relationships are actuallyestablished.
The focal points are structured on four levels: (1) firm size and organisationalstructuring of production; (2) firms’ training policies; (3) adoption of HRM practices;and (4) firm performance. Each focal point groups a set of variables that caninfluence the firm’s propensity to train, thereby determining the establishmentof complementary relationships. In this respect, each variable, either positivelyor negatively, affects the returns from training, establishing a relation ofcomplementarity or substitutability with training practices.
1. Firm’s structural variables. As stated in the theoretical section, a positive associationis expected between training provision and firm size. In addition, sector, used asa proxy for capital equipment, is expected to affect the firm’s choice in relation totraining. The influence of capital equipment is appraised using data on plantflexibility. In addition, the adoption of innovative organisational practices affectingproduction throughout each phase, such as Total Quality Management (TQM) andjust-in-time, and the fixing of the number of hierarchical layers, can affect thereturns to skill development, encouraging the financing of training.
2. Firm’s training policies. The returns from training in terms of employees’ skills areself-reinforcing. Therefore, it would be expected that there would becomplementary relationships among training practices involving differentoccupational groups. Data availability allows investigation of the hypothesis ofcomplementary relationships between training practices involving newly hiredand existing employees. The empirical analysis is developed by means of amultivariate probit.8 The data available allow us to control for the employees’educational level and the incidence of manual workers.
3. HRM practices. The empirical section of the article deals with the effect ofinnovative HRM practices such as task rotation, quality circles, teamwork,employee involvement in firm management and labour flexibility. The linkbetween training and labour flexibility9 is not unambiguous, as discussed in thetheoretical section. In terms of the other practices listed, there is no reason tobelieve that any of them are associated via a specific form of training. The
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 295
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
combination of size, internal labour market features and HRM can give rise to aframework of complementary relationships, which makes the analysis of theeffects on training of each individual practice extremely problematic. Furthermore,as pointed out in the management literature (Ricart and Portales, 2001), differentpractices may hinder one another. Consequently, these practices may substitute forand not just reinforce one another. For this reason, we cannot expect an a prioriassociation between training and HRM to hold.
4. Variables measuring firm performance. The relation between training and indicatorsof economic performance can be complex. Provision of training gives rise toincreases in both labour productivity and firm profitability. An increase inprofitability in its turn favours the accumulation of resources aimed at thefinancing of training for employees. For this reason, it is difficult to point to acausal link between any measure of training provision and any indicator of firmperformance. In any case, a positive association between these two variables isfirmly rooted in any approach to analysing training in firms. Problems arise when,based on the empirical evidence, this association turns out to be weak or, evenworse, non-existent. In this case, the only sensible conclusion is that firms do notbenefit from training, i.e. training is irrelevant to their economic performance. Thiswould then mean that if training is being provided, it is basically being used asa tool for improving the match between the characteristics of the workforce andthe organisation of work, and not in order to strengthen and widen the range ofskills used in production. Failure to find a meaningful association betweentraining and performance might indicate a poor quality of labour demand. Ofcourse, this is a rather sweeping conclusion, which should be supported by furtherempirical evidence.
Econometric analysis
The analysis of firms’ training efforts based on the dataset of companies in ReggioEmilia begins with two preliminary probit regressions concerning the provision of(formal and informal) training to existing employees and to newly hired workers.This preliminary analysis highlights the positive role played by size (medium–largefirms in the Reggio Emilia survey), labour flexibility and the adoption oforganisational innovation. The roles of past productivity performance (a real accountdata indicator) and process innovation adoptions are less important, although stillsignificant.
Table 3 presents two training indexes: the index for the variety of trainingpractices10 adopted (TRAIN-ADOP) by firms and the index related to formal trainingcoverage (TRAIN-COV). We also investigate the correlation between differenttraining practices (here, employees and newly hired) by means of a bivariate probitmodel, which specifies a joint distribution.
With respect to the four focal points discussed previously, the main conclusionsdrawn from the empirical analysis are as follows.
1. The three regressions show that there is a clear size effect. In addition, the plantflexibility and hierarchical layers indicators are negatively related to TRAIN-ADOP and TRAIN-COV. Finally, TQM positively affects the propensity to train. InReggio Emilia, TQM practices are widespread in large and even in medium-sized
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007296
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
firms, the latter being also characterised by a high degree of organisationcomplexity and facing a fierce market competition in product innovation.
2. The analysis of the complementarity of skill development among differentsegments of employees is investigated by means of a bivariate probit. There is astatistically significant relation of complementarity between the training of newlyhired and the other employees, which is shown in the last two columns of Table 3.The null hypothesis of no correlation is rejected. Results show that size factors aremore important for the training of newly hired employees than for the training ofexisting employees. Moreover, while the education/skill of the workforce explainsboth forms of training, labour flexibility is only crucial in explaining employeetraining. Organisational innovation has an influence on both forms, although
TABLE 3 Training regressions
TRAIN-ADOP TRAIN-COV TRAIN-EMPTRAIN-NEW
Newly hired EmployeeCons- 0.325 -0.980§ -2.37*** -3.14§MEDIUM 0.123** 0.104§ 0.773*** 0.132MED-LARGE 0.259§ . . . 1.045* 1.478*LARGE 0.186*** -0.076* . . . . . .COOP 0.2873§ 0.243*** . . . . . .NAT-REV . . . 0.001* . . . . . .MKT-REV . . . -0.208*** . . . . . .EDUC . . . 1.368§ 6.906*** 5.92***MANUAL 1.34§ 0.628*** . . . . . .HYERARC -1.569§ -0.624** . . . . . .PLANT-FLEX -1.306§ -0.707§ . . . . . .LAB-FLEX . . . . . . 2.048 5.816§INVOLV . . . 0.114* . . . . . .INNO-ORG § . . . ** **TEAM -0.0214 0.056 -0.739 -0.150QC 0.080* 0.045 0.353 0.872JIT 0.100* 0.022 -0.658 0.633TASK 0.035 -0.005 0.599* -0.162TQM 0.148*** 0.118*** 0.291 0.801***INNOPROC . . . 0.170§ . . . . . .PRODUC . . . 0.465§ . . . . . .correlation value
(bivariate probit). . . . . . 0,655§
F test (significance level) 0.00017 0.0000 0,0000Adj-R2 0.1638 0.266 . . .N 166 166 166
Note: We emphasise coefficients that are significant at 20, 10, 5 and 1 per cent (*, **, *** and §), derivingfrom the backwards ‘from general to particular’ estimation procedure. Pavitt sector indicators areincluded as control dummies (not shown).
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 297
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
statistically, it is not highly significant. Among HRM practices, task rotation is theonly practice affecting newly hired employees and TQM is the only significantfactor in employee regressions.
3. The analysis shows that there is a weak statistical link between the indicators fortraining and HRM practices, such as teamwork, task rotation and quality circles.Those practices generally act as a complement to training activities. We also notethat an in-depth analysis of practices may lead to less clear-cut results. Thequestion of whether it is meaningful to consider specific individual effects, or ajoint index of the intensity of higher-performance practices to capture the mainrelationships, remains open. We also generated HRM interaction variables,grouping practices in bundles of two and three. Interaction variables take thevalue 1 when two/three practices are adopted. The estimated coefficients showthat the sign of the relationship between these variables and training differs.Therefore, the effect of HRM on training depends on the specific practices adoptedand, especially, on how these combine.11 In terms of labour flexibility, its complexrelation with training is deserving of a separate analysis. As a preliminary, labourflexibility is captured in this analysis by a compound index with various elements,including numerical and functional flexibility. This indicator exerts a positiveimpact on training, although it is significant only in the bivariate probit analysis.Presumably, as was the case for HRM practices, this compound indicator hasdivergent effects that tend to offset each other.
4. Past productivity is a positive driver for training (highly significant in explainingcoverage) while net profits, although linked to positive coefficients, never reachstatistical significance. Performance indicators do not influence the ‘intensity’ oftraining practices adoption (Storey, 2004).
CONCLUSIONS AND POLICY ISSUES
This article has addressed the notion of complementarity in production as ananalytical tool for investigating the determinants of training in firms. The basic ideais that training is provided if the development of the skills to which it gives riseallows the establishment of idiosyncratic complementary relationships with otherinputs of production, as well as with structural factors, observable at firm level. Theempirical analysis provides evidence that some structural factors, such as firm sizeand TQM practices, are positively correlated with the propensity of firms to trainemployees. Furthermore, the applied analysis highlights that complementaryrelationships can be established between the training of different segments of theemployed workforce, i.e. newly hired and other employees.
These results are not particularly surprising; however, the analysis highlightscertain factors that deserve specific discussion and should be of interest to bothpolicy makers and HRM practitioners.
First, when innovative HRM practices such as task rotation, quality circles,teamwork and employee involvement are considered separately, they do not seem tobe associated with the firm’s propensity to train employees, which implies that thesepractices do not give rise to relevant complementary relationships with skillsdevelopment. A possible explanation for the low skills content of these practices isthat their implementation requires minimum on-the-job training, which employers
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007298
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
can find difficult to perceive and measure. For instance, job rotation can involve arecombination of individuals’ task allocation and does not involve further training ifemployees already know how to accomplish these tasks and if their dexterity can beimproved by a simple repetition of such tasks. In this case, the employees’endowment of skills acquired is sufficiently developed to cope with any problemsthat may arise from a recombination of tasks. In terms of quality circles, these arefrequently set up on an informal basis, out of remunerated working time, and henceare not perceived by employers as training. However, when these and similarpractices are considered in ‘bundles’, the evidence seems to point to possible positiverelations between these bundles and training.
Second, the link between training and productivity is not clear-cut, as indicatorsof profitability do not reach an evident level of statistical significance as indicatorsof past productivity.
Considering HRM separately or grouped in bundles leads to differentconclusions:
1. These two results are compatible if one assumes that these HRM practices increaselabour productivity,12 favouring the application of greater pressure and theheightening of efforts on employees, with no significant change in trainingpolicies. Intensification of efforts connected with the introduction of HRMpractices is deserving of further analysis, but at the same time, these results seemconsistent with the hypothesis of ‘management by stress’ discussed by Parker andSlaughter (1988) and outlined in Ramsay et al. (2000). In these labour processapproaches, such practices do not directly lead to an upgrading of employees’skills; they result in increased labour productivity, with no sizeable change inemployees’ skills. Subsequently, the rise in labour productivity leaves room for anintensification of training activities. Financial variables, including profits, do nothave an impact, maybe highlighting a mismanagement by firms at a dynamiclevel. The role played by (past) productivity levels suggests the establishmentof a virtuous circle based on the dynamic relationship between HRM →productivity → training → productivity.
2. The conclusions that can be drawn based on the bundling of HRM practices seemconsistent with the High-Involvement Management model outlined by Ramsayet al. (2000). They suggest that these practices give rise directly to the upgradingof employees’ skills. As a result of training, labour productivity augments and canfavour the financing of further training activities. A dynamic virtuous circle is setin motion, characterised by co-evolutionary increases in productivity and trainingefforts, probably financed mainly by sources external to the firm, indicated by thelack of a statistically significant association between profitability and training.The dynamic relation at work is represented by: HRM → training →productivity → training.
In summary, the conclusions that can be drawn from the labour process approachapply when HRM practices are separately taken account of, whereas analysis basedon the High-Involvement Management model seems more appropriate when thesame practices are adopted jointly. The effect of HRM practices on employeesdepends on the specific use made of them at firm level. This result parallels those
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 299
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
obtained in the economics literature on the link between productivity and HRM butdiffers in its emphasis on training activities rather than firm productivity.
Unlike HRM practices, policies affecting the firm as a system (Baron and Kreps,1999) seem to have an evident effect on training. In fact, TQM, labour and plantflexibility, and a decrease in the number of hierarchical levels, all of which affect thecomprehensive operation of a firm’s activities, have a substantial effect on training.These practices activate complementary relationships, whose exploitation requiresthe development of employees’ skills. The negative sign of the parameter related toplant flexibility indicates that complementarities may well result from the adjustmentof capital equipment to the existing bundle of skills.
The analysis enables some key questions on labour policy to be addressed. Trainingactivities emerge as being positively associated with firm productivity, workforce skilllevel and structural variables. The relevance of both structural variables and labourproductivity as drivers of training shows that regional industrial policies must supportlabour policies within an integrated policy effort aimed at increasing potential firmproductivity. The analysis also suggests that policies targeting the intensification oflabour flexibility should proceed hand in hand with policies providing incentives forthe adoption of structured training policies at firm level. The positive associationbetween the compound indicator of labour flexibility and training is evidence thatflexibility can be a useful tool for job creation only if training creates the conditionsnecessary for the establishment of those complementary relationships needed to makeflexible staffing convenient. This is a key concern for the current debate on localpolicies in the European and Italian environment.
Notes
1. Research for this article was carried out within the 2003–2004 PRIN ResearchProgramme ‘Structural Dynamics: Firms, Organizations and Institutions’, researchunity of Ferrara, and within the 2005–2007 local research project ‘Innovativedynamics in the knowledge-based economy. Economic analyses of open localsystems’. We acknowledge the comprehensive and valuable support of the CGILof Reggio Emilia regarding data collection at many stages of the investigation. Wethank two anonymous referees for their valuable comments. We are also indebtedto many colleagues who read the article and commented on it during informalmeetings and national and international conferences.2. See Michie and Sheehan-Quinn (2001) for a thorough analysis of labourflexibility and its effects on the firm’s performance and the firm’s propensity toinnovate. See also Kleinknecht (1998) for a critical appraisal of numericalflexibility.3. In order to verify if the firms’ sample, distributed by sectors and firm size, isrepresentative, a test was performed (Cochrane, 1977), which yielded tolerableresults.4. Although data are not panel based, we exploit both lagged firm performances(1998–2001) and introduction of organisational practices occurring before (onaverage the first introduction for the five elicited practices was between 1991 and1995) the elicited training activities (in their 2001 implementation state). Thus,some causality problems are mitigated even in a cross-sectional scenario. The linkbetween training and organisational practices is one with an eventual ‘causal
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007300
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
direction’, going from organisational innovation to training. We also recall that allother variables are elicited as trends over 1998–2001 besides training (2001).Concerning the present dataset, the empirical model that uses training as adependent variable is, statistically speaking, the more indicated and less affectedby endogeneity-deriving flaws.5. At least one of the five organisational practices studied in the literature onHRM is present in 67.3 per cent of firms. Among those, TQM, job rotation andteamwork are the most widespread, while quality circle and job rotation are lesscommon (around 10 per cent of firms).6. As indexes of training coverage, intensity or adoption of practices structurallyrange between 0 and 1, we deal with fractional variables, continuous within the0–1 range but limited by their nature. There is not an ‘optimal’ econometricmodel for studying fractional variables. Although OLS estimates may suffer fromdistortions, it is often possible to verify that estimates deriving from OLS andother more complex models do not differ significantly as far as coefficientabsolute and relative significances are concerned (Pindyck and Rubinfeld, 2000).This was confirmed by a preliminary analysis of our data. As our aim is not toestimate elasticities, we thus decided to use linear estimation procedures.Furthermore, given a sufficiently high number of firms not involved in trainingactivities, we check the presence of sample selection by a two-stages model(probit plus OLS) in all specifications.7. The table shows correlations for the most critical factors related to innovation,industrial relations and high-performance practices (HPP). It is worth noting that:(1) overall, besides a few cases, correlations tend to be rather small; (2) correlationsregarding organisational variables (TQM, QC, etc.), although positive as expected,are not so high to undermine regression analysis, if we take as a shared rule ofthumb a 0.30 correlation value as threshold; (3) correlation concerning technologicalinnovation and organisational innovations are also not high, and even lower thanabove; and (4) a positive significant correlation characterises the two indexes oforganisational innovation and industrial relations (workers’ involvement).Following this, the two are introduced separately in the regression. The correlationconfirms that high involvement and good-quality industrial relations are moreintense where innovation is higher. Although it is not possible to draw out acausality link, there is a lot of accumulated evidence that the two entangledelements are at the basis of better performances. Innovation and HPP spurproductivity, and they are more intense where the quality of industrialrelations is higher and where the involvement of workers is more developed.8. The bivariate probit is employed when one wants to test the hypothesis ofinterrelationship between two key variables. In other words, under the nullhypothesis that the covariance between the error terms of the two distinctregressions is zero, the bivariate probit consists of two independent regressions.If the null hypothesis is rejected, we face a joint co-determination of the twoinvestigated variables. In statistical terms, the errors of the two equations arerelated (a part of the error term is common to both).9. The composite index of labour flexibility includes information on short-termcontracts, functional flexibility and innovation in working hours. Table 2 describesall explanatory factors.
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 301
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
10. Training indexes refer to the complete set of formal and informal practices.11. Some of the two- and three-factor interactions are significant; the sign of therelationship is nevertheless not stable, indicating both complementarity andsubstitution effects with regard to HRM association with training when thosepractices are taken jointly. This is not surprising because the extent to whichHRM is exerted on training cannot be assessed ex ante on a purely theoreticalbasis. This result is similar to findings by other scholars. The literature is rich;see, for example, Osterman (1994), Ichniowski et al. (1997) and Michie andSheehan-Quinn (2001). However, we use as dependent variable data on training,whereas in the literature, the dependent variable is some measure of economic orfinancial performance.12. See note 10.
REFERENCES
Acemoglu, D. and Pischke, J.-S. (1999). ‘Beyond Becker: training in imperfect labourmarkets’. The Economic Journal, 109: February, 112–142.
Baron, J. and Kreps, D. (1999). Strategic Human Resources, New York: John Wiley & Sons.Becker, G. (1975). Human Capital: A Theoretical Analysis with Special Reference to Education,
New York: Columbia University Press.Beckmann, X. (2002). ‘Firm sponsored apprenticeship training in Germany: empirical
evidence from establishment data’. Labour, 16: 2, 287–300.Black, D., Noel, B. and Wang, Z. (1999). ‘On the job training, establishment size and firm
size: evidence for economies of scale in the production of human capital’. SouthernEconomic Journal, 66: 1, 82–100.
Boheim, R. and Booth, A. (2004). ‘Trade union presence and employer provided trainingin Britain’. Industrial Relations, 43: 3, 520–545.
Cochrane, W. (1977). Sampling Techniques, New York: John Wiley & Sons.Frazis, H., Herz, D. and Horrigan, M. (1995). ‘Employer provided training: results from
a new survey’. Monthly Labour Review, May, 3–17.Ichniowski, C., Shaw, K. and Prennushi, G. (1997). ‘The effect of Human Resource
Management on productivity: a study of steel finishing line’. American Economic Review,87: 3, 291–313.
Kleinknecht, A. (1998). ‘Is labour market flexibility harmful to innovation?’ CambridgeJournal of Economics, 22: 3, 387–396.
Lazear, E. (2003). Firm-Specific Human Capital: A Skills Weight Approach, NBER WorkingPaper no. 9769, Cambridge, MA: NBER.
Michie, J. and Sheehan, M. (2005). ‘Business strategy, human resources, labour marketflexibility and competitive advantage’. International Journal of Human ResourceManagement, 16: 445–464.
Michie, J. and Sheehan-Quinn, M. (2001). ‘Labour market flexibility, human resourcemanagement and corporate performance’. British Journal of Management, 12: 4, 287–306.
Milgrom, P. and Roberts, J. (1990). ‘The economics of modern manufacturing: technology,strategy and organization’. The American Economic Review, 80: 3, 511–528.
Milgrom, P. and Roberts, J. (1995). ‘Complementarities and fit. Strategy, structure andorganizational change in manufacturing’. Journal of Accounting & Economics, 19:179–208.
Osterman, P. (1994). ‘How common is workplace transformation and who adopts it’.Industrial and Labor Relations Review, 47: 2, 175–188.
Parker, M. and Slaughter, J. (1988). Choosing Sides: Unions and the Teams Concept, Boston:Labor Notes.
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007302
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
Pindyck, R. and Rubinfeld, D. (2000). Econometric Models and Economic Forecasts, London:McGraw-Hill Education.
Ramsay, H., Scholarios, D. and Harvey, B. (2000). ‘Employees and high-performance worksystems: testing inside the black box’. British Journal of Industrial Relations, 38: 4,501–531.
Ricart, J.E. and Portales, C. (2001). ‘Employment contracts, new organizational forms andcompetitive advantage for continuous innovation’, in J. Gual and J.E. Ricart (eds),Strategy, Organization and the Changing Nature of Work, Cheltenham: Edward Elgar.
Stevens, M. (1994). ‘A theoretical model of on-the-job training with imperfectcompetition’. Oxford Economic Papers, 46: 537–562.
Stevens, M. (1999). ‘Human capital theory and UK vocational training policy’. OxfordReview of Economic Policy, 15: 1, 16–32.
Storey, D. (2004). ‘Exploring the link, among small firms, between management trainingand firm performance: a comparison between the UK and OECD countries’.International Journal of Human Resource Management, 15: 1, 112–130.
Thurow, L. (1975). Generating Inequality, New York: Macmillan Press.Whitfield, K. (2000). ‘High performance workplaces, training and the distribution of
skills’. Industrial Relations, 39: 1, 1–25.
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 303
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
AP
PE
ND
IX
Tab
leA
.1C
orre
lati
onm
atri
x
hyer
arc
plan
t-fl
exed
ucla
bflex
man
ual
team
qcji
tta
sktq
min
no-
org
inno
-pr
odin
no-
proc
inno
-qu
alin
no-
tech
form
-ev
alin
volv
hyer
arc
1.00
0pl
ant-
flex
-0.0
011.
000
educ
-0.0
940.
058
1.00
0la
bflex
-0.0
920.
252
0.14
21.
000
man
ual
0.42
70.
904
0.01
20.
188
1.00
0te
am-0
.070
0.06
30.
080
0.01
00.
027
1.00
0qc
-0.1
390.
097
0.08
40.
109
0.02
90.
131
1.00
0jit
-0.0
820.
134
0.18
20.
078
0.08
60.
173
0.08
51.
000
task
-0.1
320.
050
-0.0
830.
105
-0.0
110.
236
0.20
80.
244
1.00
0tq
m-0
.111
0.12
50.
026
0.17
50.
066
0.13
30.
218
0.27
30.
318
1.00
0in
no-o
rg-0
.175
0.15
10.
078
0.15
90.
062
0.57
70.
487
0.54
50.
695
0.68
41.
000
inno
-pro
d-0
.018
-0.1
060.
054
-0.1
19-0
.103
-0.0
35-0
.006
-0.0
84-0
.098
-0.1
03-0
.113
1.00
0in
no-p
roc
0.08
30.
025
0.01
60.
116
0.05
80.
164
-0.0
42-0
.052
-0.0
330.
049
0.04
1-0
.316
1.00
0in
no-q
ual
0.03
10.
070
-0.2
300.
141
0.07
6-0
.010
0.03
10.
051
0.21
10.
154
0.15
6-0
.309
0.13
21.
000
inno
-tec
h0.
008
0.18
1-0
.033
0.21
40.
167
0.06
80.
124
0.07
30.
161
0.29
40.
250
-0.4
670.
591
0.64
41.
000
form
-eva
l-0
.130
-0.0
200.
039
0.06
8-0
.074
0.17
40.
145
0.19
30.
163
0.19
00.
285
-0.0
830.
031
0.02
80.
132
1.00
0in
volv
-0.1
650.
157
0.16
90.
252
0.07
10.
604
0.29
00.
261
0.29
10.
291
0.58
1-0
.139
0.12
00.
089
0.24
20.
595
1.00
0
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007304
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
TAB
LE
A.2
Tota
lfir
mpo
pula
tion
;sh
ares
byse
ctor
and
size
Sec
tor
Siz
e
50–9
910
0–24
925
0–49
950
0–99
9>9
99To
tal
Tota
l(a
bsol
ute
valu
e)
Food
and
beve
rage
s0.
78%
1.95
%1.
17%
0.78
%0.
78%
5.45
%14
Oth
erin
dus
trie
s0.
78%
0.00
%0.
00%
0.00
%0.
00%
0.78
%2
Pape
ran
dpr
inti
ng1.
56%
0.00
%1.
17%
0.00
%0.
00%
2.72
%7
Che
mic
alpr
oduc
ts,
rubb
eran
dpl
asti
cm
ater
ials
3.11
%2.
72%
0.78
%0.
00%
0.39
%7.
00%
18
Woo
d0.
00%
0.78
%0.
00%
0.00
%0.
00%
0.78
%2
Met
alpr
oduc
ts,
met
alw
orki
ngeq
uipm
ent,
tran
spor
teq
uipm
ent,
elec
tric
ald
evic
es,
mec
hani
cal
mac
hine
ry
28.0
2%15
.95%
5.06
%2.
72%
3.50
%55
.25%
142
Non
-met
alm
iner
als
(Cer
amic
)9.
73%
6.61
%1.
95%
2.72
%0.
78%
21.7
9%56
Text
ile&
clot
hing
1.56
%1.
56%
2.72
%0.
00%
0.39
%6.
23%
16To
tal
45.5
3%29
.57%
12.8
4%6.
23%
5.84
%10
0.00
%To
tal
(abs
olut
eva
lue)
117
7633
1615
257
Giovanni Guidetti and Massimiliano Mazzanti
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007 305
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.
Tab
leA
.3In
terv
iew
edfir
ms
(199
),da
taus
edfo
rec
onom
etri
can
alys
is(1
66):
shar
esby
sect
oran
dsi
zean
dsa
mpl
ete
st
Sec
tor
Siz
e
50–9
910
0–24
925
0–49
950
0–99
9>9
99To
tal
Tota
l(a
bsol
ute
valu
e)
Food
and
beve
rage
s0.
00%
60.0
0%10
0.00
%10
0.00
%10
0.00
%71
.43%
1O
ther
ind
ustr
ies
100.
00%
––
––
100.
00%
2Pa
per
and
prin
ting
75.0
0%–
100.
00%
––
85.7
1%6
Che
mic
alpr
oduc
ts,
rubb
eran
dpl
asti
cm
ater
ials
100.
00%
71.4
3%10
0.00
%–
100.
00%
88.8
9%16
Woo
d–
50.0
0%–
––
50.0
0%1
Met
alpr
oduc
ts,
met
alw
orki
ngeq
uipm
ent,
tran
spor
teq
uipm
ent,
elec
tric
ald
evic
es,
mec
hani
cal
mac
hine
ry73
.61%
73.1
7%84
.62%
85.7
1%10
0.00
%76
.76%
109
Non
-met
alm
iner
als
(Cer
amic
)68
.00%
88.2
4%10
0.00
%10
0.00
%10
0.00
%82
.14%
46Te
xtile
&cl
othi
ng75
.00%
75.0
0%28
.57%
–10
0.00
%56
.25%
9To
tal
73.5
0%75
.00%
78.7
9%93
.75%
100.
00%
77.4
3%To
tal
(abs
olut
eva
lue)
8657
2615
1519
9
Food
and
beve
rage
s0.
00%
60.0
0%10
0.00
%10
0.00
%10
0.00
%71
.43%
10O
ther
ind
ustr
ies
100.
00%
––
––
100.
00%
2Pa
per
and
prin
ting
75.0
0%–
100.
00%
––
85.7
1%6
Che
mic
alpr
oduc
ts,
rubb
eran
dpl
asti
cm
ater
ials
87.5
0%57
.14%
100.
00%
–0.
00%
72.2
2%13
Woo
d–
50.0
0%–
––
50.0
0%1
Met
alpr
oduc
ts,
met
alw
orki
ngeq
uipm
ent,
tran
spor
teq
uipm
ent,
elec
tric
ald
evic
es,
mec
hani
cal
mac
hine
ry59
.72%
68.2
9%76
.92%
71.4
3%88
.89%
66.2
%94
Non
-met
alm
iner
als
(Cer
amic
s)44
.00%
64.7
1%80
.00%
85.7
1%10
0.00
%60
.71%
34Te
xtile
&cl
othi
ng75
.00%
25.0
0%14
.29%
–10
0.00
%37
.50%
6To
tal
58.9
7%63
.16%
69.7
%81
.25%
86.6
7%64
.6%
Tota
l(a
bsol
ute
valu
e)69
4823
1313
166
Sam
ple
repr
esen
tati
vene
sste
st(C
ochr
an,
1977
)Te
stm
argi
nof
erro
r0.
045
(Cri
tica
lm
argi
nof
erro
rfo
rsm
all
sam
ples
0.10
);T
hefo
rmul
ais
:n
=N
/[(
N-
1)th
eta2
+1]
;w
here
nis
the
sam
ple
(166
),N
the
univ
erse
(257
),an
dth
eta
the
erro
rw
efa
ce
Training and organisational innovations in a local industrial system
HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 17 NO 3, 2007306
© 2007 The Authors.
Journal compilation © 2007 Blackwell Publishing Ltd.