Advanced Technology, Innovation, Wages and Productivity in The Canadian Manufacturing Sector

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Applied Economics Letters

ISSN: 1350-4851 (Print) 1466-4291 (Online) Journal homepage: http://www.tandfonline.com/loi/rael20

Advanced technology, innovation, wages and


productivity in the Canadian manufacturing sector

Brian P. Cozzarin

To cite this article: Brian P. Cozzarin (2016) Advanced technology, innovation, wages and
productivity in the Canadian manufacturing sector, Applied Economics Letters, 23:4, 243-249,
DOI: 10.1080/13504851.2015.1068913

To link to this article: http://dx.doi.org/10.1080/13504851.2015.1068913

Published online: 05 Aug 2015.

Submit your article to this journal

Article views: 19

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rael20

Download by: [Central Michigan University] Date: 09 December 2015, At: 04:30
APPLIED ECONOMICS LETTERS, 2016
VOL. 23, NO. 4, 243–249
http://dx.doi.org/10.1080/13504851.2015.1068913

Advanced technology, innovation, wages and productivity in the Canadian


manufacturing sector
Brian P. Cozzarin
Department of Management Sciences, University of Waterloo, Waterloo, Ontario, Canada

ABSTRACT KEYWORDS
In this article, we used a two-step estimation procedure, where in the first stage, the number of Product innovation; process
advanced manufacturing technologies used in the firm was estimated using a negative innovation; advanced
binomial regression, and the expenditure on process and product innovation was estimated manufacturing technology;
using a type II Tobit procedure. In the second stage, we used the predicted values from the productivity; wages
first stage in wage and labour productivity equations. The data were from the 2009 Survey of JEL CLASSIFICATION
Downloaded by [Central Michigan University] at 04:30 09 December 2015

Innovation and Business Strategy which was linked to the General Index of Financial O31; O32
Information (2004–2009) and the Longitudinal Employment Analysis Program (2004–2009).
The implications for policy are that we should not expect large aggregate effects of innovation
on productivity and employment. We should expect wage increases and productivity increases,
with process innovation. We should also expect moderate wage increases with product
innovation, and contrary to process innovation, the effect on productivity of product innova-
tion was negative.

I. Introduction community innovation surveys (CIS), Pianta and


Tancioni (2008) found that wages grow faster in
Canada’s recent budget announced nearly $10 bil-
sectors where innovation expenditure is higher.
lion targeted towards innovation or at activities that
They also discovered that factors affecting wages
support innovation (Flaherty 2010). With such funds
were different for high- and low-innovation sectors.
targeted towards innovation, it is important to quan-
In terms of productivity and innovation, Crepon,
tify the value of innovation to industry.
Duguet, and Mairessec (1998) using the French
Furthermore, the government is concerned that
Office of Industrial Studies and Statistics (SESSI)
advanced technology adoption in Canada lags the
survey found a positive effect of innovation on pro-
US (Gilroy 2013). The objectives of the article are
ductivity. In a study of Italian firms, Hall, Lotti, and
to find the impact of product and process innovation
Mairesse (2012) found that process innovation had a
and advanced technology on manufacturing wages
negative effect on productivity while product inno-
and productivity. The policy implications are as fol-
vation had a positive effect. Using CIS3 and CIS4,
lows: if the effects of innovation are positive then
Mairesse and Robin (2009) found that product inno-
government spending is justified, if the effects are
vation affects labour productivity positively but pro-
negligible then government spending is questionable
cess innovation does not. Huergo and Jaumandreu
or perhaps it should be re-directed to areas other
(2004) in a Spanish study of manufacturing firms
than innovation, perhaps to commercialization
found that process innovation caused productivity
instead. We will first discuss innovation and wages
growth.
and then innovation and productivity. In a Finnish
Here, we will discuss advanced manufacturing
study, Kalmi and Kauhanen (2008) found that wages
technology (AMT) in terms of wages and produc-
benefit from innovation. Martínez-Ros (2001) using
tivity. It has been shown that flexibility often
Spanish data found that firms with joint product-
reduces the engineering cost of design changes
process innovation pay higher wages and that inno-
and systems modifications. However, the savings
vation is a good approximation for rents. Using
are often offset by increased programming costs.

CONTACT: Brian P. Cozzarin brian.cozzarin@uwaterloo.ca


© 2015 Taylor & Francis
244 B. P. COZZARIN

There will also often be a decrease in unskilled technologies are new technologies (equipment or
labour requirements but an increase in those of software) that perform a new function or improve
skilled labour (Sanchez 1996) with higher wages. some function significantly better than commonly
Feenstra and Hanson (1999) used the NBER used technologies in the industry or by your
Productivity Database from 1979 to 1990 to inves- competitors.’ The survey lists nine AMTs: advanced
tigate computer investment and worker wages. computerized design and engineering; advanced
They found that the expenditure on computers computerized processing, fabrication and assembly;
increased relative wages for nonproduction work- advanced computerized inspection technologies;
ers. Doms, Dunne, and Troske (1997) used the advanced communication technologies; advanced
1988 Survey of Manufacturing Technology and automated material handling technologies; advanced
the Worker-Establishment Characteristic Database information integration and control technologies;
to examine the wage effect of AMT (N = 3260). advanced biotechnologies/bioproducts; advanced
They found that plants that use a large number of nanotechnologies and advanced green technologies.
new technologies pay higher wages and employ For industry dummies we adapt Pavitt’s (1984) tax-
onomy – with five classifications for North
Downloaded by [Central Michigan University] at 04:30 09 December 2015

more educated workers along with more man-


agers, professionals and precision-craft workers. American Industry Classification System (NAICS)
Dunne and Schmitz (1995) used the US 1988 manufacturing industries: labour intensive
Census Bureau Survey of Advanced Technology (N = 674), resource intensive (N = 658), scale inten-
and the 1987 Census of Manufactures sive (N = 802), science (N = 234) and specialized
(N = 6909). They found that plants that used the (N = 347).4 The remaining variable names and sta-
largest number of AMT paid the highest wages; tistics are mentioned in Table 1.
these plants also employed the largest proportion
of nonproduction workers. In terms of productiv-
ity and AMT, Bartelsman, Van Leeuwen, and III. Model estimation
Nieuwenhuijsen (1998) used statistics from the Given the forgoing discussion from the
Netherlands for 1985 and 1991 and linked to the ‘Introduction’ section, two hypotheses related to
1992 Survey of Advanced Manufacturing innovation and AMT are postulated:
Technology and to the Survey of Capital Stocks. (1) Product and process innovations increase
Variables included output, value added, employ- manufacturing wages and labour productivity.
ment, wage, productivity, K-L ratio and an AMT (2) Use of AMT increases manufacturing wages
index. The results showed that firms using AMT and labour productivity.
in 1992 showed higher average growth rates of We use a modified ‘CDM’ model that employs two
productivity between 1985 and 1991. sequential stages (Mairesse and Robin 2009).
Econometric specifications for the product innova-
tion, process innovation and advanced technology
II. Survey of Innovation and Business Strategy
equations are as follows. Specify in the first stage a
(SIBS) and linkage to GIFI and LEAP
type II Tobit model, where the probit part of the
The SIBS 20091 was linked with a unique firm iden- selection process estimates the probability of under-
tifier to the General Index of Financial Information taking process (product) innovation, and the maxi-
(GIFI)2 collected by Canada Revenue Agency and mum likelihood OLS part estimates the expenditure
the Longitudinal Employment Analysis Program on process (product) innovation. Also in the first
(LEAP).3 The final sample size after linking and stage is the estimation of the number of AMT used
cleaning is 2715 firms. For our purposes, we use by the firm using a negative binomial estimator. The
the AMT definition from SIBS: ‘advanced predicted values from the type II Tobit expenditure
1
The survey was conducted in January‒April 2010 by Statistics Canada under the auspices of the 1985 Statistics Act which legally requires firms to answer
the questionnaire. The sample was 6233 firms with at least 20 employees; response rate was 70%. For more information, see http://www.ic.gc.ca/eic/site/
eas-aes.nsf/eng/h_ra02092.html.
2
For more information, see http://www.cra-arc.gc.ca/tx/bsnss/tpcs/crprtns/rtrn/wht/gifi-ogrf/menu-eng.html.
3
For more information, see http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=8013.
4
Due to space limitations, the NAICS4 concordance with the taxonomy is omitted.
APPLIED ECONOMICS LETTERS 245

Table 1. Regression variables. proc exp ¼ μ2 numamt þ β2 X1 þ ε2


0
(2)
Variable Description (source) Mean SE
dequity* 2007–2009 change in equity (GIFI) −0.2301 0.0452 First-stage type II Tobit for product innovation and
demp* 2007–2009 change in employment −0.1705 0.0420
(LEAP)
expenditure:
dwage* 2007–2009 change in wage (LEAP) −0.1510 0.0593 0
dyl* 2007–2009 change in Y/L (GIFI, −0.1990 0.0672 Prðprod binÞ ¼ μ3 numamt þ β3 X1 þ ε3 (3)
LEAP) 0
dkl* 2007–2009 change in K/L (GIFI, −0.0687 0.0626 prod exp ¼ μ4 numamt þ β4 X1 þ ε4 (4)
LEAP)
proc_exp 2009 process innovation 330,846 61,589 First-stage negative binomial for AMT:
expenditure (Q49)
prod_exp 2009 product innovation 477,640 87,866 0
expenditure (Q88) numamt ¼ β5 X2 þ ε5 (5)
numamt 2009 number of AMT in use, up to 9 1.6596 0.0510
(Q44) Second-stage performance equations using predicted
proc_bin Binary = 1 if had process innovation 0.1730 0.0200
(2007–2009) values from stage one:
prod_bin Binary = 1 if had product 0.1810 0.0150
innovation (2007–2009) ^
dwage ¼ φ1 demp þ φ2 numamt þ φ3 proc^ exp
DE2006 Debt to equity ratio 2006 (GIFI) −1859.58 1895.72 (6)
þ φ prod^ exp þβ X3 þ ε6
0
Downloaded by [Central Michigan University] at 04:30 09 December 2015

DE2007 Debt to equity ratio 2007 (GIFI) 31.77 11.26


4 6
DE2008 Debt to equity ratio 2008 (GIFI) 22.20 10.01
aXpc Interaction: number of AMT, (189,731) 37,823 ^
dyl ¼ ψ1 dkl þ ψ2 numamt þψ3 proc^ exp
expenditure process innovation (7)
þ ψ prod^ exp þβ X3 þ ε7
0
aXpd Interaction: number of AMT, (1,438,042) 107,151
4 7
expenditure product innovation
pcXpd Interaction: process innovation 4.36E+13 1.17E+12 where dwage is 2007–2009 change in wage, demp is
expenditure, product innovation
expenditure 2007–2009 change in employment, dyl is 2007–2009
hrm HR practices (standardized variable −0.1828 0.0259 change in output to labour ratio (Y/L = labour pro-
from 9 items in Q64)
university Percent employees with university 0.1071 0.0036 ductivity) and dkl is 2007–2009 change in capital to
degree (Q63)
change_bp Change to business practices −0.0075 0.0288
labour ratio (K/L). By using differenced variables,
(standardized variable from 9 items we eliminate unobserved fixed effects for employ-
in Q38)
ob_export Obstacles to exporting −0.1292 0.0278 ment, wages, K/L and Y/L. All variables with ‘^’ are
(standardized variable from 12 predicted values from the first-stage equations.
items in Q37)
hoffice Binary = 1 if head office is Canada 0.8927 0.0069
(Q10)
outsource Outsourcing activities (standardized 0.0748 0.0069 Independent variables
variable from 14 items in Q24)
Note: HR, human resource; *** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05. The average debt to equity ratio will be used in the
innovation equations (Mairesse et al. 1999). Chen
and Huang (2009) found a positive relationship for
equations and the AMT equation are then used in hiring practices, employee participation and perfor-
stage 3 – wage and productivity equations. mance appraisals for administrative innovation,
We will assume that technology adoption is endo- while for technical innovation staffing, participa-
genous (Chennells and VanReenan 1997, 1998; Siegel tion and compensation were significant. Question
1998). And we will assume that product and process 64 in SIBS lists nine HR practices that the employer
innovation are endogenous. This explains why we use can checkmark related to the following: aptitude
all three predicted variables in the wage and produc- tests for selection of new employees, formal train-
tivity equations. The error structure in the wage ing programmes to do their job, formal training
equation and the productivity equation is unknown programmes to aid in promotion, formal perfor-
(since we are combining predicted values from nega- mance agreements, formal appraisals for nonman-
tive binomial estimator and a two-step type II Tobit agement staff, formal appraisals for management,
procedure) and there may be bias from the predicted incentive programmes to nonmanagement staff,
regressors; thus, we use bootstrapped SEs. incentive programmes to management and incen-
First-stage type II Tobit for process innovation tive programmes that are available to all employees.
and expenditure: We created a normalized variable called hrm for
0
‘human resource management’ from these prac-
Prðproc binÞ ¼ μ1 numamt þ β1 X1 þ ε1 (1) tices. From question 63, we created university
246 B. P. COZZARIN

which is the percentage of employees with a uni- Table 2. Negative binomial regression for number of AMT.
versity degree. numamt Coef. SE
DE2006 4.95E-07 2.49E-06
Business practices can impact innovation and DE2007 7.68E-06 0.000
thus SIBS includes nine items related to changing DE2008 −2.27E-05 0.000
hrm 0.190*** 0.023
business practices in question 38. A standardized university 0.382** 0.141
variable was created called change_bp from the fol- change_bp 0.260*** 0.022
hoffice −0.027 0.057
lowing items: cost reduction, improved good/ser- outsource 0.062 0.068
vice quality, decreased lead times, increased after- ob_export 0.125*** 0.022
b_lab −0.193** 0.073
sales functions, acceptance of greater risk sharing, b_res −0.242*** 0.074
incurred greater up-front or nonrecurring costs, b_sca −0.141* 0.070
b_sci 0.084 0.091
entered a new geographic region or expanded exist- _cons 0.594*** 0.077
N 2715
ing operations, undertook a new business activity LR χ2(13) 378.3
and extended business hours to accommodate Prob > χ2 0.0000
Log likelihood −4739.4
employees/customers/suppliers. From question 37 Pseudo-R2 0.04
(a nonforced Likert scale for ‘low’, ‘medium’ and
Downloaded by [Central Michigan University] at 04:30 09 December 2015

Note: *** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05.


‘high’), we created a normalized variable called
ob_export for obstacles to exporting. The obstacles
listed in question 37 include legal/administrative, to equity ratio of the firm had no effect on the
export taxes, uncertainty about international stan- number of AMT in use.
dards, access to financing, concerns of violating Table 3 shows the type II Tobit procedure for
patents/intellectual property rights, foreign tariffs/ expenditure on process innovation. The first-stage
trade barriers, border security issues, distance to probit equation has a binary dependent variable
customers, linguistic/cultural issues, customer
requirements to use specific technology/systems, Table 3. Type II Tobit model for expenditure on process
meeting cost requirements and meeting quality innovation.
requirements. Outsourcing activities of the firm Type II Tobit selection model
(coded as ‘yes’/‘no’) are reported in question 24. A Coef. SE
standardized variable was created using the 14 proc_exp Step 2: Maximum likelihood estimation
numamt 7,39,430*** 80,932
items: production of goods, provision of services, dequity −12,363 61,842
distribution and logistics, call centres, marketing/ hrm 7,48,391*** 1,65,841
university 4,21,791 1,029,418
sales/after sales service, software development, data change_bp 1,121,531*** 1,64,849
hoffice 39,583 4,04,010
processing, ICT (information and communications outsource 1,678,886*** 4,97,421
technology) services, legal services, accounting/ ob_export 3,71,734* 1,58,713
b_lab 0.440 206
book-keeping, human resource management b_res 0.243 174
(HRM), financial management, engineering/technical b_sca 0.647 189
b_sci 0.835 272
services and R&D. Finally, a binary variable called _cons −3,951,208 4,28,493
hoffice was created that equals 1 if the firm has its proc_bin Step 1: Probit for sample selection
head office in Canada. numamt 0.108*** 0.012
dequity −0.002 0.009
hrm 0.109*** 0.024
university 0.062 0.150
IV. Estimation results change_bp 0.164*** 0.024
hoffice 0.006 0.059
outsource 0.245*** 0.073
The first stage estimates for the number of AMT and ob_export 0.054* 0.023
expenditure on process and product innovation are _cons −0.577*** 0.063
N 2515
displayed in Tables 2–4. For the negative binomial Censored obs. 1054
regression of number of AMT (Table 2), we see that Uncensored obs. 1461
Wald χ2(12) 283.7
human resource practices, change in business prac- Prob > χ2 0.0000
Log likelihood −25581.5
tices and obstacles to exports are all highly signifi- LR test for independence χ2(1) = 1825.2 Prob > χ2 = 0.0000
cant. The proportion of university educated of equations
employees is significant at the 1% level. The debt Note: *** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05.
APPLIED ECONOMICS LETTERS 247

Table 4. Type II Tobit model for expenditure on product Table 5. Bootstrap OLS performance equations.
innovation. dwage dyl
Type II Tobit selection model (change in (change in labour
Coef. SE wage productivity
2007–09) SE 2007–09) SE.
prod_exp Step 2: Maximum likelihood estimation
numamt 1,436,311*** 2,34,905 demp 1.187*** 0.043
dequity −3,02,785 1,85,007 dkl 0.771*** 0.084
hrm 1,683,465*** 4,82,652 ahat# 0.130 1.176 −0.020 4.747
university 1.33E+07*** 2,873,255 pchat# 7.85E-07*** 1.470E-07 1.57E-06** 5.550E-07
change_bp 3,240,482*** 4,80,338 pdhat# −4.15E-07*** 7.690E-08 −7.86E-07** 2.790E-07
hoffice 1,119,526 1,171,739 aXpc## −6.25E-10 4.830E-08 −1.09E-07 1.230E-07
outsource 4,764,503*** 1,430,169 aXpd## 6.72E-09 1.630E-08 3.76E-08 4.860E-08
ob_export 1,524,812*** 4,62,289 pcXpd## −4.46E-16 1.480E-15 −9.06E-16 4.270E-15
b_lab −1.105 584.075 hrm 0.081 0.225 0.169 0.896
b_res −1.291 593.808 university 5.067*** 1.018 9.065* 3.931
b_sca −0.767 556.287 change_bp 0.405 0.314 0.746 1.242
b_sci 0.843 741.340 hoffice 0.386*** 0.090 0.587* 0.277
_cons −14,700,000*** 1,242,935 outsource 0.641*** 0.133 1.076* 0.484
ob_export 0.352* 0.157 0.615 0.616
prod_bin Step 1: Probit for sample selection b_lab 0.048 0.230 −0.103 0.912
numamt 0.0739*** 0.0121 b_res −0.002 0.290 −0.041 1.140
−0.0156
Downloaded by [Central Michigan University] at 04:30 09 December 2015

dequity 0.0095 b_sca 0.016 0.170 −0.159 0.667


hrm 0.0867*** 0.0248 b_sci −0.091 0.113 −0.037 0.457
university 0.6854*** 0.1483 _cons −2.854*** 0.887 −4.996 3.447
change_bp 0.1668*** 0.0246 N 2293 2293
hoffice 0.0576 0.0604 Bootstrap 500 500
outsource 0.2453*** 0.0736 replications
ob_export 0.0785*** 0.0238 R2 0.90 0.59
_cons −0.7558*** 0.0638 Root MSE 0.54 1.44
N 2515
Censored obs. 1226 Notes: #indicates a predicted variable from either type II Tobit estimation
Uncensored obs. 1289 (pchat, pdhat) or from negative binomial estimation (ahat).
##
Wald χ2(12) 238.5 indicates an interaction variable created using predicted values from type
Prob > χ2 0.0000 II Tobit and negative binomial estimation (aXpc, aXpd) or from type II
Log likelihood −24027.7 Tobit estimation (pcXpd).
LR test for independence χ2(1) = 1923.1 Prob > χ2 = 0.0000 *** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05.
of equations
Note: *** p ≤ 0.001; ** p ≤ 0.01; * p ≤ 0.05.
interaction terms are also insignificant. We see that
both pchat (predicted value of expenditure on process
called proc_bin, while the second-stage OLS regres- innovations) and pdhat (predicted value of expendi-
sion has process innovation expenditure as the depen- tures on product innovation) are significant. In the
dent variable. We see that HRM intensity, intensity of wage equation, the elasticity for pchat is 12.60 and
changes to business processes, outsourcing and obsta- 5.59 for pdhat. So for a 1% increase in the expenditure
cles to exporting are substantially significant. Table 4 on process innovation, there is a 12.60% change in
shows the type II Tobit procedure for engaging in and wages. However, as the table shows, product and pro-
expenditure on product innovation. The same vari- cess innovation were found to impact wages and
ables as in Table 3 are significant; however, two more labour productivity. The elasticity for process innova-
variables arise: change in equity and the proportion of tion in the wage equation is 12.6% while for product
university-educated employees. This indicates that innovation it is 5.6%. This demonstrates that wages are
increased cash flow and a higher proportion of uni- more affected by process innovation. In the labour
versity-educated employees have a significant impact productivity equation, process innovations have an
on the likelihood of conducting product innovation elasticity of 0.4% and product innovations have an
and on the level of expenditure. elasticity of −6.8%.
Table 5 shows the results from the bootstrap regres- What can be said to generalize these findings?
sions for change in wage and change in labour produc- We see that process innovation increases wages.
tivity. From the table, we see that ahat is insignificant Pianta and Tancioni (2008) used European CIS
in both regressions. Our finding of no effect of AMT is and found that wages grew faster in sectors where
in contrast to the literature (Bartel and Sicherman innovation expenditure was higher. Unfortunately,
1999; Bartelsman, Van Leeuwen, and Nieuwenhuijsen our results cannot corroborate their findings. In the
1998; Doms, Dunne, and Troske 1997; Dunne and productivity equation process, innovation reduces
Schmitz 1995; Hitt and Brynjolfsson 1996). The productivity.
248 B. P. COZZARIN

V. Conclusion Chennells, L., and J. Van Reenen. 1998. “Establishment Level


Earnings, Technology and the Growth of Inequality:
In this article, we used a modified ‘CDM’ two-stage Evidence from Britain.” Economics of Innovation and
estimation procedure. In the first stage, we estimated New Technology 5 (2–4): 139–164. doi:10.1080/
the number of AMT, expenditure on product and 10438599800000003.
process innovation; in the second stage, the predictions Chennells, L., and J. Van Reenan. 1997. “Technical Change
and Earnings in British Establishments.” Economica 64
from stage one were used in wage and productivity
(256): 587–604. doi:10.1111/ecca.1997.64.issue-256.
equations. Data were from the SIBS which was linked Crepon, B., E. Duguet, and J. Mairessec. 1998. “Research,
to the General Index of Financial Information and the Innovation and Productivity: An Econometric Analysis at
LEAP. In the wage equation, the elasticity for process the Firm Level.” Economics of Innovation and New
innovation is 12.6% and for product innovation it is Technology 7 (2): 115–158. doi:10.1080/10438599800000031.
5.6%. In the labour productivity equation, process Doms, M., T. Dunne, and K. R. Troske. 1997. “Workers, Wages
and Technology.” The Quarterly Journal of Economics 112
innovations have an elasticity of 0.4% and product
(1): 253–290. doi:10.1162/003355397555181.
innovations have an elasticity of −6.8%. The implica- Dunne, T., and J. A. Schmitz. 1995. “Wages, Employment
tions for policy are that we should not expect large Structure and Employer Size-Wage Premia: Their
Downloaded by [Central Michigan University] at 04:30 09 December 2015

aggregate effects of innovation on productivity. Relationship to Advanced Technology Usage at US


However, we should expect wage increases and pro- Manufacturing Establishments.” Economica 62: 89–107.
ductivity increases, with process innovation. We Feenstra, R. C., and G. H. Hanson. 1999. “The Impact of
Outsourcing and High-Technology Capital on Wages:
should also expect moderate wage increases with pro-
Estimates for the United States, 1979–1990.” The Quarterly
duct innovation, but should not expect productivity to Journal of Economics 114 (3): 907–940. doi:10.1162/
increase. 003355399556179.
Flaherty, J. M. 2010. Budget 2010: Leading the Way on Jobs
and Growth. Ottawa, ON: Department of Finance, Her
Disclosure statement
Majesty the Queen in Right of Canada.
No potential conflict of interest was reported by the author. Gilroy, G. 2013. Evaluation of the Digital Technology
Adoption Pilot Program Final Report. Ottawa, ON:
National Research Council of Canada.
Funding Hall, B. H., F. Lotti, & J. Mairesse. 2012. “Evidence on the
Impact of R&D and ICT Investments on Innovation and
This work was supported in part by Social Sciences and Productivity in Italian Firms.” Economics of Innovation
Humanities Research Council under Grant 435-2014-2061 and New Technology 22 (3): 300–328. doi:10.1080/
and by Industry Canada under Research Contract 5025591.
10438599.2012.708134.
Hitt, L. M., and E. Brynjolfsson. 1996. “Productivity,
Business Profitability, and Consumer Surplus: Three
ORCID Different Measures of.” MIS Quarterly 20 (2): 121–142.
Huergo, E., and J. Jaumandreu. 2004. “Firms’ Age, Process
Brian P. Cozzarin http://orcid.org/0000-0002-4280-7633
Innovation and Productivity Growth.” International
Journal of Industrial Organization 22: 541–559.
References doi:10.1016/j.ijindorg.2003.12.002.
Kalmi, P., and A. Kauhanen. 2008. “Workplace Innovations
Bartel, A. P., and N. Sicherman. 1999. “Technological and Employee Outcomes: Evidence from Finland.”
Change and Wages: An Interdindustry Analysis.” Journal Industrial Relations 47 (3): 430–459.
of Political Economy 107 (2): 285–325. Mairesse, J., B. Hall, L. Branstetter, and B. Crepon. 1999.
Bartelsman, E., G. Van Leeuwen, and H. Nieuwenhuijsen. “Does Cash Flow Cause Investment and R&D: An
1998. “Adoption of Advanced Manufacturing Technology Exploration Using Panel Data for French, Japanese,
and Firm Performance in the Netherlands.” Economics of and United States Scientific Firms.” In Innovation,
Innovation and New Technology 6 (4): 291–312. Industry Evolution, and Employment, edited by D.
doi:10.1080/10438599800000023. Audretsch, and A. R. Thurik. Cambridge: Cambridge
Chen, C.-J., and J.-W. Huang. 2009. “Strategic Human University Press.
Resource Practices and Innovation Performance ‒ The Mairesse, J., and S. Robin. 2009. Innovation and Productivity:
Mediating Role of Knowledge Management Capacity.” A Firm-Level Analysis for French Manufacturing and
Journal of Business Research 62 (1): 104–114. Services Using CIS3 and CIS4 Data. Malakoff: Centre de
doi:10.1016/j.jbusres.2007.11.016. Recherche en Economie et Statistiques.
APPLIED ECONOMICS LETTERS 249

Martínez-Ros, E. 2001. “Wages and Innovations in Spanish Sanchez, A. M. 1996. “Adopting Advanced Manufacturing
Manufacturing Firms.” Applied Economics 33 (1): 81–89. Technologies: Experience from Spain.” Journal of
doi:10.1080/00036840150203251. Manufacturing Systems 15 (2): 133–140. doi:10.1016/
Pavitt, K. 1984. “Sectoral Patterns of Technical Change: 0278-6125(96)82338-7.
Towards a Taxonomy and a Theory.” Research Policy 13 Siegel, D. S. 1998. “The Impact of Technological Change on
(6): 343–373. doi:10.1016/0048-7333(84)90018-0. Employment: Evidence from a Firm-Level Survey of Long
Pianta, M., and M. Tancioni. 2008. “Innovations, Wages, and Island Manufacturers.” Economics of Innovation and New
Profits.” Journal of Post Keynesian Economics 31 (1): Technology 5 (2–4): 227–246. doi:10.1080/104385998000
101–123. doi:10.2753/PKE0160-3477310105. 00006.
Downloaded by [Central Michigan University] at 04:30 09 December 2015

You might also like