ARTICLE IN PRESS

0925-5273/$ - see

doi:10.1016/j.ijp

�Correspondifax: +1785 532

E-mail addre

clyang@chu.edu

Int. J. Production Economics 103 (2006) 863–872

www.elsevier.com/locate/ijpe

Productivity of production labor, non-production labor,and capital: An international study

John G. Wackera, Chen-Lung Yangb, Chwen Sheuc,�

aDepartment of Management, Arizona State University, Tempe, AZ 85287-4006, USAbDepartment of Technology Management, Chung-Hua University, Hsin-Chu 300, Taiwan

cDepartment of Management, Kansas State University, Manhattan, KS 66506, USA

Received 1 May 2004; accepted 1 December 2005

Available online 9 June 2006

Abstract

Productivity is defined as the amount of output produced with certain combinations of input resources (capital, labor,

etc.). Recent studies have indicated the value of non-production labor (e.g., engineers, product designer, quality inspectors,

and administrators) to a manufacturing plant’s productivity. However, the effect of non-production labor compared to

other input resources such as production labor and capital on factory productivity has not been fully investigated. Without

understanding how individual input resources affect productivity, manufacturing firms can mismanage resource

investment, which will ultimately hinder the growth of productivity. This study examines the relative effect of input

resources on factory productivity across countries. We use data collected from 508 manufacturing plants in 16 countries to

estimate and compare productivity of input resources between countries. Statistical results are presented and directions for

future research are suggested.

r 2006 Elsevier B.V. All rights reserved.

Keywords: International management; Manufacturing strategy; Factory productivity; Non-production labor; International manufacturing

1. Introduction

Productivity is an index that measures outputrelative to input. Plants with higher productivityproduce more output for a given level of input thanplants with lower productivity. This higher produc-tivity results in lower input levels to produce thesame good or service, giving the firm a potentialcompetitive advantage in the international market-place (Lowe and Fernandes, 1994; Grubbstrom and

front matter r 2006 Elsevier B.V. All rights reserved

e.2005.12.012

ng author. Tel.: +1785 532 4363;

7024.

sses: john.wacker@asu.edu (J.G. Wacker),

.tw (C.-L. Yang), csheu@ksu.edu (C. Sheu).

Olhager, 1997; Mefford, 1991). Due to the signifi-cance of this issue, many researchers in variousdisciplines (e.g., economics, operations manage-ment, and engineering) have continuously studiedthe subject of productivity. There are variousresearch issues pertaining to productivity includingproductivity measures, factors that affect produc-tivity, office vs. factory productivity, and ways toimprove productivity (Stevenson, 2004).

Productivity is defined as the amount of outputproduced with certain combinations of inputresources (e.g., capital, labor, etc.). While therecan be many possible input resources, labor andcapital have been the two primary input resourcesconsidered in most productivity research in the

.

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872864

fields of economics and operations management.More recently, many studies have begun to discussthe value of non-production workers (i.e., manage-rial, technical, and support staff such as engineers,product designers, quality inspectors, purchasingmanagers, and administrators) to a manufacturingplant’s productivity (Gray and Jurison, 1995;Gunasekaran et al., 1994; Kang and Hong, 2002;Krajewski and Ritzman, 2004). As automationtechnology replaces traditional workers, the pro-ductivity of non-production workers relative toother input resources becomes critical to improvingfactory productivity. The relative contribution ofdifferent input resources to productivity is an evenmore significant issue from the perspective ofinternational manufacturing and outsourcing.Facing the trend of globalization, many multi-nationals have been investing in overseas facilities toimprove international competitiveness. Since differ-ent countries can have over- or under-investment ofdifferent input resources (Corvers, 1997), it is likelythat managers in different countries must manageresources differently to improve plant productivity.Unfortunately, very few studies have specificallyexamined the international productivity of manu-facturing plants on a large scale. The relative impactof various input resources (including non-produc-tion labor) on productivity between countries hasnot been studied despite the increasing establish-ment of foreign facilities. The lack of understandingof the management of plant productivity in differentcountries can mislead resource investment andultimately hinder the growth of productivity.

This study examines the relative influence ofvarious input resources on manufacturing plantproductivity across countries. We use data collectedfrom 508 manufacturing plants in 16 countries toperform empirical analysis. The following sectionreviews relevant studies of international productiv-ity followed by a discussion of research methodol-ogy. Both input and output measures ofproductivity are suggested. The statistical resultsand discussion are presented, followed by sugges-tions for future research.

2. Literature review

The trend of increasing contributions to produc-tivity from non-production workers definitely pre-sents a new challenge to productivity research andpractice that traditionally include only productionworkers (Gray and Jurison, 1995; Krajewski and

Ritzman, 2004). While no studies have specificallyexamined the productivity of non-production work-ers, a few researchers have empirically observed thevalue of non-production activities. For instance,Hayes and Clark (1986) investigated how organiza-tions used different strategic resources to improveproductivity. They analyzed 12 factories of threecompanies over time, and their findings indicatedthat productivity increase was associated withcapital investment in new equipment; reductions inmaterial waste, work-in-process inventory, and thenumber of engineering change orders, and increasedemployee learning rates. Accordingly, they sug-gested that investment in both human and equip-ment resources improved plant productivity.Schmenner (1991) analyzed factories in NorthAmerica (USA and Canada), Europe (France,Germany and England), and Korea. His resultsrevealed that investment in new equipment andincreased worker participation increased the pro-ductivity of the plants. Lieberman (1989) foundmanufacturing productivity increased from variousnon-production activities such as well-defined tasks,employee improvement suggestions, and interactionbetween production employees and equipment/tooling engineers. A stream of research has alsosuggested that various improvement activities per-formed under the names of quality managementand environmental management increased plantproductivity (Flynn et al., 1995; Karmarkar andPitblanddo, 1997; Klassen and Whybark, 1999;Sousa and Voss, 2001). Finally, the concept ofsupply chain management has verified the contribu-tion of non-production activities to productivity.

While researchers have begun to recognize thecontribution of non-production workers to produc-tivity, none have investigated the relative effect ofnon-production workers and other input resourceson plant productivity. Flaherty (1996) comparedmanufacturing firms in Singapore, Mexico, andTaiwan and reported the differences in their‘‘sensitivity’’ of productivity growth to labor andcapital investment. No explanation was offered forwhy such difference occurs. Corvers (1997) investi-gated variables affecting labor productivity withinthe EU member states. His results showed over-investment of human capital in some countries andunder-investment in other countries. Yamada et al.(1997) studied the influence of labor, capital, andR&D on productivity and found a higher produc-tivity contribution from capital resources (measuredas value-added).

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872 865

Overall, the common thread woven throughprevious studies is that non-production employeesare important for improving the material flow,product design, and process design. Ultimately theseimprovements increase productivity. However, therelative impact of various input resources, especiallynon-production workers, is not fully understood.Moreover, there is a lack of international produc-tivity studies at the plant level regarding the relativecontribution of various input resources despite thegrowing number of foreign facilities. Not knowinghow to manage input resources to increase produc-tivity can definitely result in waste of resourceinvestment. This study extends previous studies byexamining the relative impact of several inputresources on manufacturing plant productivitywithin a country and compares those results acrosscountries. To make this comparison, this studyincorporates three input resources: productionlabor, non-production labor (e.g., engineers, pro-duct designers, quality inspectors, purchasing man-agers, administrators, etc.), and capital equipment.Therefore, the objectives of this study are to addressthe following research questions:

(1)

What is the relative effect of three inputresources of production (production labor,non-production labor, and capital equipment)on productivity?

(2)

Does the relative effect of individual productioninput resources vary by country?

3. Research methodology—production function

3.1. Cobb– Douglas production function

The mathematical description of productivity interms of the amount of output produced withcertain combination of input resources (capital,labor, etc.) is as follows:

q ¼ f ðv1; . . . ; vnÞ, (1)

where q represents output level and (v1; . . . ; vn)stands for n input resources. While there are manyalternative forms of the production function, theCobb–Douglas (CD) function has been verifiedempirically and is suggested as a valid descriptionof reality by studies conducted in different nationsand industries (Corvers, 1997; Hsing, 1993; Jia,1991; Maurice and Thomas, 1995; Yamada et al.,1997). For example, Hsing (1993) compared five

different production functions and concluded thatthe CD function was appropriate for both durableand non-durable industries. Other production func-tion forms such as the constant elasticity ofsubstitution and trans-log functions seem to havedifficulty with empirical estimation of these impacts(Arrow et al., 1961; Hsing, 1993). In addition, theCD production function can be easily modified forexamining the impact of specific individual inputresources (Corvers, 1997; Yamada et al., 1997; Tay,1992). Therefore, this study will use the CD functionfor the productivity analysis. The following is ageneralized CD function:

q ¼ Aðv1Þaðv2Þ

b, (2)

where A, a, and b are constants, and q is the outputlevel generated by inputs v1 (capital) and v2 (labor).The assumption of this function is that a significantdegree of output changes can be explained byvariations in the capital/labor ratio. In this study,labor is divided into two categories to examine thedifferential impact on output resulting from pro-duction workers vs. the non-production workers. Amodified form of the CD production function isused. The exact form to be used is as follows:

Output ¼ b0ðLaborPÞbPðLaborNÞbNðCapitalÞbKe,

(3)

where Output ¼ output level per year; La-borP ¼ total production worker labor hours peryear; LaborN ¼ total non-production worker laborhours per year; Capital ¼ total new productionequipment investment per year; and b0, bP, bN,and bK are the parameters to be estimated and e isthe stochastic error.

In this study, value-added is used as the measurefor output. The traditional estimating procedurecalls for a double-logarithmic transformation of theexplanatory variables and the output-dependentvariable. Therefore, the specific estimation formfollowing this transformation is

LnðOutputÞ ¼ b0 þ bP LnðLaborPÞ þ bN LnðLaborNÞ

þ bK LnðCapitalÞ þ LnðeÞ. ð4Þ

The estimated coefficients (bP, bN, and bK) arepartial output elasticities with respect to productionlabor, non-production labor, and capital, respec-tively. These partial elasticities are defined as theratio of the percentage change in output tothe percentage change in input. In other words, ifthe input resource i increases by 1%, output willincrease by bi percent. A particular input resource

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872866

with higher elasticity is referred to as havingrelatively larger impact on output level. Therefore,elasticity is used to gauge the impact of inputresources on productivity. All output elasticities areunit free, which alleviates the problems of dataincomparability between countries. The summationof the elasticities for all three independent variables(production labor, non-production labor, and capi-tal investment) is an estimate of returns-to-scale (seeJorgenson, 1986 for a detailed explanation). If thissummation is greater than one, there are increasingreturns-to-scale (meaning a doubling of all inputswill produce more than a doubling of outputs), anda summation of less than one means that there aredecreasing returns-to-scale (doubling all inputs willproduce less than double output).

Note that the estimates of bP, bN, and bK can beused to compare the elasticity of each input resourcefor any country to those of any other country.However, the estimates cannot be used to assess theabsolute levels of productivity for each inputresource. Assessing and comparing the absolutelevels of productivity for each input resourcerequires productivity measures that are adjusted toaccount for differences between countries in ex-change rates, inflation by industry, and pricedifferentials. Because of this adjustment difficulty,most productivity data, including that reported bythe US Bureau of Labor Statistics, report onlypercent changes in productivity without compar-isons of productivity levels between countries. Thisresearch focuses on the relative effect of variousinput resources of production within a country;therefore, a comparison of the absolute levels ofproductivity for each input resource betweencountries is unnecessary.

While the concept of the production function isstraightforward, choosing appropriate input andoutput measures to develop the function is difficult.The rest of this section discusses the input andoutput measures selected to estimate the productionfunction in Eq. (4).

3.2. Choice of input and output measures

If all factories produced a homogenous product,output could be measured by counting the physicalquantity of products produced. Unfortunately, inthe physical world, most factories produce widelydifferent products, with different qualities, andother traits. There are no physical measures thatcan be used to compare output across most

manufacturing plants. As a result, gross measuressuch as total sales revenues are often used as outputmeasures. Unfortunately, using sales as a measureof output distorts the amount of output a plantproduces, since it does not account for differences inmaterial purchases. For example, one company mayhave very few internal operations but still have highsales output due to large material purchases, whileanother plant may have a greater number ofinternal operations but have considerably lowersales due to little material usage. Compared to salesvolume, value-added has been a more recognizedoutput measure in literature (Corvers, 1997; Hsing,1993; Jia, 1991; Tay, 1992; Yamada et al., 1997).This study uses value-added (sales minus costs ofpurchased material) as the output measure to studyproductivity at the plant level.

For input measures, the number of labor hoursworked is generally used to measure the two laborresources: production and non-production workers(Krelle, 1983). Nevertheless, the choice of themeasure for the third input resource, capital, ismore controversial. Existing literature suggests twowidely used measures of capital investment: bookvalue of equipment (purchase price minus deprecia-tion), and new capital investment. In practice,neither can be easily measured without possiblebias (Yamada et al., 1997). For book value, thereare many different ways that depreciation can becalculated, and it would be an impossible task tocompare different accounting standards betweencountries. Furthermore, the book value of equip-ment does not necessarily represent the amount ofcapital equipment actually used as an input, sincesome equipment may be sitting idle yet still have anaccounting book value (Hildebrand and Liu, 1965).

Recent studies have supported the use of newcapital investment to represent the capital input, butno general agreement has been reached (e.g.,Yamada et al., 1997). The problem with using newequipment investment as the input measure is thatnew capital is generally not utilized strictly asreplacement for the consumption of capital equip-ment. Consequently, new capital investment wouldnot correspond to the amount of equipmentexpended in the current period but rather to somelater period. Despite this potential problem, newcapital investment may still be a better representa-tion of the actual amount of equipment consumedfor cross-country comparisons. From a theoreticalperspective, new equipment investment appears tobe the best alternative to measure capital input.

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872 867

Appendix A includes the relevant survey questionsrelated to both input and output measures.

This study performed two statistical tests todetermine whether the book value of equipment orthe yearly investment in new equipment should beused as a proxy for capital input. The first test usedregression analysis to check for the statistical fit ofthe production function (Eq. (4)) to a similarequation with the book value of equipment sub-stituted for new equipment investment. The analysisfound that for 14 of the 16 countries, the productionfunction with new equipment investment had higherexplanatory power than the equation using the bookvalue of equipment. We also performed stepwiseregressions for each country with the book value ofequipment as a fourth independent variable in Eq. (4).The book value of equipment significantly improvedprediction for only two of the 16 countries, Japan andRussia. In both cases, new equipment investment wasbetter than the book value of equipment in predictingvalue-added. Adding the book value of equipment tothe regression equation improved R2 by less than 2%.Thus, some empirical justification exists in this dataset for using new equipment investment as a proxy forcapital input.

3.3. Data collection

The data were collected by the Global Manufac-turing Research Group (GMRG) in an extensivedata-gathering effort throughout the world. GMRGis a multi-national community of researchersdedicated to the study of international operationsmanagement (Whybark, 1997). Its primary goal isto promote an understanding of differences inmanufacturing practices across internationalboundaries through joint research efforts. TheGMRG data were primarily collected from twoindustries: non-fashion textiles and machine tools.Survey questions covered areas of manufacturingactivities such as sales forecasting, productionplanning and scheduling, shop floor control, pur-chasing and materials management, and manufac-turing performance. This survey questionnaire hasbeen previously validated in many studies publishedin Decisions Sciences, Journal of Operations Man-

agement, International Journal of Production Re-

search, etc. Full details about its developmentand the administration of the survey are availablein Whybark and Vastag (1994) and Whybark(1997).

4. Statistical results

Table 1 presents the regression estimates of thetransformed CD production function. As indicatedby the F-ratio and the R2 statistic, all estimatedregression equations are highly significant.

With the exception of Northern Ireland, in all 16countries at least one of the three input resources(production labor, non-production labor, and capi-tal equipment) is related to output in the CDproduction function estimates. Several tests wereperformed to determine whether multi-collinearityhad an impact on the results in Table 1. The first testexamined the variance inflation factor (VIF) foreach variable for each regression equation. The VIFindicates the degree to which the variance of theestimated coefficient is overstated due to collinearitywith other variables in the regression equation. Thisoverstatement can possibly produce a non-signifi-cant t-test for a coefficient that would be significanthad collinearity not been present. The mostcommon ‘‘rule of thumb’’ is that the VIF shouldnot exceed 10 (Freund and Wilson, 1997). Thisexamination revealed that Northern Ireland hadVIF statistics above 10 (14.45 and 15.191) forproduction and non-production worker coefficients,respectively.

There are several traditional remedies for multi-collinearity. One method is to drop the offendingvariable. However, this is not recommended whenthere are strong theoretical reasons for including thevariables. Therefore, a stepwise regression wasperformed to determine the source of multi-colli-nearity. In this estimate, the data from NorthernIreland revealed that non-production workers hadthe highest output elasticity (t ¼ 8:67, po0:01).Accordingly, we conclude that multi-collinearitydoes not bias the estimated coefficients (Greene,1990).

We also performed a Chow test to address theconcern of combining data from two distinctindustries into a single estimate. The resultsindicated there were no significant differencesbetween the regression coefficients for the twoindustries (po0:05). This finding is supported byprevious empirical evidence (Schmenner, 1991).

The fifth column of Table 1 computes the returns-to-scale ðbP þ bN þ bKÞ for all 16 countries. Theresults reveal that no country had return-to-scaleestimates significantly different from 1.0. In otherwords, there are no statistical reasons to believe thatreturns-to-scale are increasing or decreasing for any

ARTICLE IN PRESS

Table 1

Cobb–Douglas production function parameter estimates

(1) Country (2) bP (Production

labor)

(3) bN (Non-

production labor)

(4) bK (Capital) (5)

Returns-

to-scale

(6) R2 (7) F-ratio (8)

Sample

size

Bulgaria 0.625* (0.021) 0.101 (0.355) 0.069 (0.377) 0.794 0.643 6.599 15

1,126.5 a 326.5 US$ 3.6M b (0.008 c)

China 0.157 (0.132) 0.692* (0.035) 0.199* (0.016) 1.049 0.938 20.166 8

893.4 761.8 US$ 2.9 M (0.000)

England 0.466* (0.047) 0.490** (0.018) 0.041 (0.402) 0.998 0.807 13.854 14

158.8 113.2 US$ 4.6 M (0.001)

Germany 0.793** (0.004) 0.427** (0.021) 0.031 (0.432) 1.251 0.757 120.20 14

245.4 198.3 US$ 11.9 M (0.000)

Hungary 0.262 (0.061) 0.442** (0.000) 0.330**(0.000) 1.034 0.770 52.363 51

685.9 478.7 US$ 9.8 M (0.000)

Ireland 0.402 (0.162) 0.502 (0.117) 0.314*(0.045) 1.219 0.786 12.269 14

362.5 212.6 US$ 10.3 M (0.001)

Japan 0.054 (0.075) 0.265** (0.001) 0.638**(0.000) 0.957 0.955 182.166 57

265.4 107.9 US$ 23.9 M (0.000)

Mexico 0.155 (0.388) 0.329 (0.252) 0.504**(0.000) 0.988 0.950 125.397 30

629.2 259.2 US$ 14.9 M (0.000)

New Zealand 0.395 (0.100) 0.615** (0.005) 0.071 (0.315) 1.081 0.813 13.063 13

180.9 60.2 US$ 0.8 M (0.000)

Northern Ireland 0.469 (0.213) 0.578 (0.136) 0.037 (0.427) 1.084 0.905 22.105 11

300.9 152.6 US$ 7.8 M (0.001)

Poland 0.301 (0.078) 0.271** (0.050) 0.198* (0.030) 0.770 0.697 16.075 25

539.9 358.5 US$ 1.4 M (0.000)

Russia 0.278 (0.058) 0.145 (0.209) 0.522**(0.000) 0.945 0.855 79.544 43

2,233.6 421.4 US$ 12.5 M (0.000)

Spain -0.207 (0.209) 0.195 (0.278) 0.840**(0.000) 0.828 0.815 42.524 33

425.1 162.0 US$ 21.4 M (0.000)

Sweden 0.700** (0.001) 0.276* (0.017) 0.148 (0.067) 1.124 0.960 11.859 (0.000) 18

141.3 84.7 US$ 4.73 M

USA -0.098 (0.160) 0.227** (0.005) 0.825**(0.000) 0.954 0.839 189.817

(0.000)

113

199.0 141.6 US$ 17.6 M

Wales 0.272* (0.017) 0.490** (0.000) 0.281**(0.001) 1.043 0.855 88.294 49

202.1 105.7 US$ 4.7 M (0.000)

Weighted Averages 0.18262 0.32059 0.47726 0.98047 0.84221 Total: 508

Note: a1000 h; bUS$ 000,000; cp-value; **p-valueso0.01; *p-valueso0.05.

J.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872868

country in this sample. The managerial implicationis that there is no appreciable productivity gain forincreasing the size of the current average plant interms of workforce or capital equipment. Theremaining section discusses the results with regardto the two research questions, elasticities ofindividual production resources and between-coun-try differences.

4.1. Elasticities of individual production inputs

The results (Columns 2–4 in Table 1) indicate thatelasticities of individual production inputs withineach country are significantly different. Overall, theelasticities of non-production and capital equipment

are significant (po0:05) in more countries thanthose of production workers. Capital and non-production labor both had 10 estimates that werestatistically significant, while only five of theproduction labor estimates were significant. In otherwords, for most countries capital equipment or non-production workers have greater impact on pro-ductivity than production workers do. The findingaddresses the second research question that therelative effect of individual input resources varies bycountry. More analysis and discussion of between-country differences are presented in the next section.

To investigate the overall effect of individualinput resources (Research question #1), we com-puted the weighted averages of the coefficients of

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872 869

these three production resources across all countriesand the results are given in the bottom row ofTable 1. The weighted averages are 0.18262,0.32059, and 0.47726 for production employees,non-production employees, and capital, respec-tively. The implication is that changes to productionworker hours, on average, have less influence onoverall output as compared to changes in non-production worker hours and capital. Capital hasthe largest impact on the productivity consistentwith results found by other researchers (Arrowet al., 1961).

An additional CD production function wasestimated using the aggregate data from all coun-tries. In this overall estimate, the production workercoefficient was statistically significant (po0:01) witha negative sign. Other previous studies have alsofound negative coefficients for production labor(Moroney, 1972; Yamada et al., 1997). The negativecoefficient of production workers indicates thatproduction workers might not contribute to factoryproductivity in comparison with the other two inputresources. The negative contributions may beattributed to automation that decreases the overallneed for and impact of direct production labor orother factors such as the increases in engineeringefforts toward process improvement. Regardless ofthe reasons for negative coefficients of productionlabor variable, this finding is important for manu-facturing managers who endeavor to increase plantproductivity.

Table 2

Ranking of relative elasticities

Country Most productive

resource

bP (Production

labor)

Germany Production 1

Sweden Production 2

Bulgaria Production 3

Poland Production 8

China Non-production 12

New Zealand Non-production 7

Northern Ireland Non-production 4

Ireland Non-production 6

England Non-production 5

Wales Non-production 10

Hungary Non-production 11

Spain Capital 16

USA Capital 15

Japan Capital 14

Russia Capital 9

Mexico Capital 13

4.2. Between-country elasticity differences

Between-country differences in productivity gen-erally can be attributed to broad sweeping areassuch as social, cultural, political, or economicdifferences. However, these broad areas offer littleinsight into the underlying causes for these produc-tivity differences at the plant level. This studyattempts to use the elasticity results to explain whysome countries have similar productivity patterns.

Table 2 sorts all countries by the ‘‘most produc-tive resource’’ and produces three groups, produc-tion, non-production, and capital. An interestingfinding is that the two countries with highestproduction employee elasticities (i.e., Germanyand Sweden) have higher returns-to-scale as com-pared to other countries. Since both countries areEuropean, this high elasticity may explain whyproduction workers in European countries have ahigher status than in other countries as Hayes andWheelwright (1984) observed.

On the other hand, some of the British influencedareas (Ireland, Northern Ireland, New Zealand,Wales, and England) (Deresky, 1997) appear tohave higher elasticity for non-production employeesthan either capital investment or production labor.In many international management studies, Ireland,Northern Ireland, New Zealand, Wales, and Eng-land are combined because they exhibit the samecultural tendencies (e.g., Ronen and Shenkar, 1985).For instance, these plants in these countries and

bN (Non-

production labor)

bK (Capital) Returns-to-scale

8 16 1.251

10 11 1.124

16 13 0.794

11 10 0.770

1 9 1.049

2 12 1.081

3 15 1.084

4 7 1.219

5 14 0.998

6 8 1.043

7 6 1.034

14 1 0.828

13 2 0.954

12 3 0.957

15 4 0.945

9 5 0.988

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872870

China may have production systems that are morebureaucratic or centrally controlled, where non-production employees are required to controlproduction through supervising workers. Balasu-bramanyam and Fu (2003) reported the critical roleof non-production and supervisory workers inChinese manufacturing plants. Effective adminis-tration is critical to the productivity of manymedium-sized enterprises. In addition, engineeringlabor may play a significant role in these plants byimplementing process improvements to improveoutput. A few researchers have confirmed the effectof national culture on manufacturing decisions (e.g.,Pagell et al., 2005), but how and why cultureinfluences plant productivity has not been empiri-cally studied and would be an important researchissue for future investigation.

The countries with the highest capital elasticitiesare Spain (1), USA (2), Japan (3), and Russia (4).The high output elasticity with respect to investmentmeans that, in these countries, capital equipment isrelatively scarce since they are higher up on themarginal productivity curve (demand for labor).The interpretation of these results suggests threeresearch hypotheses. First, the types of productsproduced by these countries may be more techno-logically advanced and therefore, cause a greaterneed for more capital to produce these products.Next, these countries may rely more heavily onadvanced manufacturing technologies for produc-tivity, thereby making the elasticity of capital higher(i.e., those countries that use more robots have moreautomated plants where one worker can operatemultiple machines). Finally, it is possible that thesecountries focused on high value-added activities,which requires more capital equipment. Overall,these three hypotheses support Schmenner’s (1991)and Hayes and Clark’s (1986) research findings thatdeveloped countries must increase capital intensityif they are to increase their productivity to remaincompetitive in international markets.

Unfortunately, these interpretations may explainUSA and Japan’s relatively higher elasticity of capitalinvestment, but they do not support the results forRussia and Spain. Future research may look forpossible explanations for the high capital elasticity forRussia and Spain based on the theory of relativescarcity of capital (Goldman et al., 1994). Namely, it ispossible that both Russia and Spain have relatively lowamounts of funds available for investment in equip-ment, which causes them to be in dire need of morecapital equipment. Overall, the between-country differ-

ences in elasticities of input resources present manyinteresting research hypotheses for future testing.

5. Conclusions

Recent studies have indicated the effect of non-production labor on a manufacturing plant’sproductivity. However, no large-scale empiricalstudies have been conducted to verify and comparesuch effects with other input resources such asproduction labor and capital. Moreover, researchon input resource productivity at the factory level islacking in international management. Assessing andcomparing the overall levels of factory productivityacross nations is extremely difficult due to differ-ences in exchange rates, inflation, trade policies, andprice differentials. This study used 508 samplescollected from 16 countries to examine the relativeelasticities of three input resources (productionworkers, non-production workers, and capitalequipment) on factory productivity, without theneed to estimate and compare relative productivitylevels. In addition to confirming the productivity ofnon-production workers and relative effect of threeinput resources on productivity, we also developedthe following hypotheses:

1.

In general, production workers have lowerelasticity than non-production workers or capitalequipment in most countries studied. A possibleinference of these findings is that the currentfactories are more automated and need non-production workers to support output.

2.

Investment in capital equipment has a majorimpact on output in developed countries. Thisfinding supports early research evidence (Hayesand Clark, 1986; van Ark and Pilat, 1993;Schmenner, 1991; Yamada et al., 1997) thatimproved capital investment could lead toimproved productivity.

3.

Central European countries and those heavilyinfluenced by the British Empire have higherlevels of non-production employee elasticity thanmost other countries. Future research is neces-sary to identify causes.

Unfortunately, limited by the current data set, weare not able to explain all the statistical results.Moreover, many issues related to between-countrycomparisons could not be fully addressed in thisstudy without including additional factors, such asnational culture and resource scarcity. We have

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872 871

proposed several research hypotheses, which must beverified or refuted in the future. Obviously, this studyis a crude starting point for international operationsmanagement studies to better understand the differ-ences between countries as to how their manufactur-ing plants may improve and invest in their resources.Finally, this study investigated the relative effect ofthe input resources that have an impact on outputcaused by changes in the individual factors ofproduction. Future studies of international factoryproductivity should adjust the relative price andexchange rate levels between countries to comparethe base for each input resource and to make actualcountry comparisons of the level of productivity ofeach input resource. In general, the understanding ofhow and why factory productivity varies betweencountries is still a largely untapped area of research.As the overseas outsourcing and ventures continue togrow, this research area becomes more critical forinternational operations management.

Appendix A. Relevant survey questions

Input measures: direct and indirect labor hours

1.

How many employees work for the company? 2. How many of these employees are direct and

indirect production workers?

3. How many hours per year does a production

employee typically work?

Input measure: new capital investment

1.

What were the company’s total sales? 2. On average, over the last 2 years, about what

percent of annual sales has been invested in newmanufacturing equipment?

Output measure: value-added

1.

What were the company’s total sales? 2. About what percent of the company’s sales is the

total manufacturing cost?

3. About what percent of the company’s total

manufacturing cost is for material?

References

Arrow, K., Chenery, H.B., Minhas, B., Solow, R.M., 1961.

Capital–labor substitution and economic efficiency. Review of

Economics and Statistics 43 (3), 225–250.

Balasubramanyam, V.N., Fu, X., 2003. Township and village

enterprises in China. Journal of Development Studies 39 (4),

27–46.

Corvers, F., 1997. The impact of human capital on labour

productivity in manufacturing sectors of the European

Union. Applied Economics 29, 975–987.

Deresky, H., 1997. International Management. Addison-Wesley,

Reading, MA.

Flaherty, M.T., 1996. Global Operations Management.

McGraw-Hill, New York.

Flynn, B.B., Schroeder, R.G., Sakakibara, S., 1995. The impact

of quality management practices on performance and

competitive advantage. Decision Sciences 26 (5), 659–691.

Freund, R.F., Wilson, W.J., 1997. Statistical Methods, Revised

Ed. Academic Press, New York.

Goldman, M.I., von Rooy, J.P., Harken, R.R., Nicandros, C.S.,

Topp, K.M., Rosenbloom, A.H., Kvint, V., 1994. The

Russian investment dilemma. Harvard Business Review

June–July, pp. 35–44.

Gray, P., Jurison, J., 1995. Productivity in the Office and the

Factory. International Thomson Publishing, Denver, CO.

Greene, W.H., 1990. Econometric Analysis, second ed. Prentice-

Hall, Englewood Cliffs, NJ.

Grubbstrom, R.W., Olhager, J., 1997. Production and flexibility:

Fundamental relations between two major properties and

performance measures of the production system. Interna-

tional Journal of Production Economics 36 (2), 169–183.

Gunasekaran, A., Korukonda, A.R., Virtanen, I., Yli-Olli, P.,

1994. Improving productivity and quality in manufacturing

organizations. International Journal of Production Econom-

ics 52 (1/2), 73–82.

Hayes, R.H., Clark, S.C., 1986. Why some factories are more

productive than others? Harvard Business Review 64, 66–73.

Hayes, R.H., Wheelwright, S.C., 1984. Restoring Our Competi-

tive Edge. Wiley, New York.

Hildebrand, G.H., Liu, T., 1965. Manufacturing Production

Functions in the United States: An Interindustry and

Interstate Comparison of Productivity. Cornell University,

New York.

Hsing, Y., 1993. On the choice of production functions: The case

of US manufacturing industries. Applied Economics 25 (3),

321–324.

Jia, L., 1991. A quantitative analysis of Chinese industrial

structure and technological change: Production functions for

aggregate industry, sectoral industries, and small scale

industry. Applied Economics 23 (11), 1733–1740.

Jorgenson, D.W., 1986. In: Grilches, Z., Intrilligator, M.D.

(Eds.), Econometric Methods for Modeling Producer Beha-

vior. Handbook of Econometrics, vol. 3. North-Holland,

New York, pp. 1841–1915.

Kang, S., Hong, D.P., 2002. Technological changes and demand

for skills in developing countries. The Developing Economics

2, 188–207.

Karmarkar, U., Pitblanddo, R., 1997. Quality, class, and

competition. Management Science 43 (1), 27–39.

Klassen, R.D., Whybark, D.C., 1999. The impact of environ-

mental technologies on manufacturing performance. Acad-

emy of Management Review 42 (6), 599–615.

Krajewski, L.J., Ritzman, L.P., 2004. Operations Management:

Strategy and Analysis. Prentice Hall, Upper Saddle River, IL.

Krelle, W., 1983. The aggregate production function and the

representative firm. In: Adams, F.G., Hickman, B.G. (Eds.),

ARTICLE IN PRESSJ.G. Wacker et al. / Int. J. Production Economics 103 (2006) 863–872872

Global Econometrics. The MIT Press, Cambridge, MA,

pp. 175–187.

Lieberman, M.B., 1989. Learning, productivity, and US–Japan

industrial competitiveness. In: Ferdows, K. (Ed.), Managing

International Manufacturing. North-Holland, New York.

Lowe, P., Fernandes, E., 1994. The growth and convergence of

manufacturing productivity in industrial and newly indus-

trialising countries. International Journal of Production

Economics 34 (2), 139–149.

Maurice, S.C., Thomas, C.F., 1995. Managerial Economics.

Richard Irwin, Boston, MA.

Mefford, R.N., 1991. Quality and productivity: The linkage.

International Journal of Production Economics 42 (1/2), 137–145.

Moroney, J.R., 1972. The Structure of Production in American

Manufacturing. The University of North Carolina Press,

Chapel Hill, NC.

Pagell, M., Katz, J., Sheu, C., 2005. The importance of national

culture in operations management: An exploratory study of

decision-making. International Journal of Operations and

Production Management 25 (4), 371–394.

Ronen, S., Shenkar, O., 1985. Clustering countries on attitudinal

dimensions: A review and synthesis. Academy of Manage-

ment Review 10 (3), 435–454.

Schmenner, R.W., 1991. International factory productivity

gains. Journal of Operations Management 10 (2),

229–254.

Sousa, R., Voss, C.A., 2001. Quality management: Universal or

context dependent? Production and Operations Management

20, 91–109.

Stevenson, W.J., 2004. Operations Management. McGraw-Hill,

Boston, MA.

Tay, R., 1992. Automation and production efficiency: An

empirical analysis of ten manufacturing industries in a newly

industrialized economy. International Journal of Technology

Management 7 (4/5), 314–327.

van Ark, B., Pilat, D., 1993. Productivity levels in Germany,

Japan, and the United States: Differences and causes.

Microeconomics 2, 1–69.

Whybark, D.C., 1997. GMRG survey research in operations

management. International Journal of Operations and

Production and Management 17 (7), 690–700.

Whybark, D.C., Vastag, G. (Eds.), 1994. Global Manufacturing

Practice: Collected Papers. Elsevier, New York.

Yamada, T., Yamada, T., Liu, G.G., 1997. Value-added

productivity of manufacturing industries in Japan. Applied

Economics 29, 1317–1324.