1、The role of performance measures and incentive systems inrelation to the degree of JIT implementationRosemary R. Fullertona, Cheryl S. McWattersb,*aUtahStateUniversity,SchoolofAccountancy,Logan,UT84322-3540,USAbFacultyofManagement,McGillUniversity,SamuelBronfmanBuilding,1001SherbrookeStreetWest,Mont
2、real,CanadaH3A1G5AbstractThe shift to world-class manufacturing strategies has necessitated complementary changes in management account-ing systems (MAS). Using survey data obtained from top manufacturing executives at 253 US rms, this studyempirically examines the relationship between the level of
3、just-in-time (JIT) practices implemented by US manu-facturing rms and the performance measures and incentive systems that are incorporated in their MAS. The statisticaltests provide empirical evidence that the use of non-traditional performance measures such as bottom-up measures,product quality, an
4、d vendor quality, as well as incentive systems of employee empowerment and compensation rewardsfor quality production are related to the degree of JIT practices implemented. # 2002 Elsevier Science Ltd. All rightsreserved.1. IntroductionAn appropriate organizational structure, incor-porating the man
5、agement accounting system(MAS), is considered a necessity for the successfulimplementation of organizational strategy (Brick-ley, Smith, Oldham Safayeni, Purdy, Van Engelen, Young White White, Pearson, White White Foster Hall Kalagnanam yetthe manufacturing practices of ecient materialow, improved q
6、uality, and increased employmentinvolvement continue to be sought-after, compe-titive strengths of world-class manufacturingrms. JIT remains the most universally acceptedterm to describe this broad production system(White Gilbert, 1990; Goyal Snell Wafa White Hedin Hiromoto, 1991; Howell Johnson Maz
7、achek, 1993;McNair, Lynch, McNair, Mos-coni, 1989; Neely, 1999; Sillince Wisner Ittner Johnson Kaplan, 1983; Kaplan Ansari Barney,1986; Bennett Hendricks, 1994;Milgrom Banker, Potter, Durden, Hassel, Ittner Jazayeri Patell, 1987). World-class manufacturing systemsadvanced manu-facturing technology,
8、TQM, and JIT have beenreferred to and examined in research studies inisolation and as synergistic combinations (Flynn,Sakakibara, Patell, 1987; SimSnell Swanson Mazachek,1993).Thecurrentstudy extends existing research related to world-class manufacturing practices and their relation-ship to performa
9、nce measures by examining thelinkagebetweenthedegreeofJITimplementationinUS manufacturing rms and the use of non-tradi-tionalperformancemeasuresandincentivesystems.2.3. Hypothesis1ATo make decisions in a JIT environment, a rmmust measure and report those items that areaected by JIT adoption (i.e. in
10、ventory turns,714 R.R.Fullerton,C.S.McWatters/Accounting,OrganizationsandSociety27(2002)711735delivery time, scrap, quality, setup times, andvendor performance). Young (1992) points outthat without appropriate measures to evaluate andcontrol the critical measures of success in a JITsystem, its level
11、 of performance could be incor-rectly assessed. According to Foster and Horngren(1987), JIT rms depend less on nancial measuresand more on personal observations and non-nancial measures. In their 1991 study of Japanesecompanies, Daniel and Reitsperger found thatsetup times, scrap, and downtime were
12、reportedmore frequently to managers supporting zero-defect strategies than managers supporting moretraditional strategies. Banker et al. (1993b)concluded that when quality improvement strate-gies were implemented, non-nancial informationto workers was more available. Results in a rela-ted study by B
13、anker et al. (1993a) indicated thatthe availability and use of productivity measureswere related to the implementation of JIT andTQM.Ittner and Larcker (1995, 1997b) recommendedthat the distribution of information encompass alllevels of the organization to overcome the de-ciencies of the MAS. Worker
14、s need to gather theirown bottom-up information using statisticalprocess control (SPC), Pareto analysis, histo-grams, and ow charts, rather than be dependentupon top-down information that emphasizesstandards and budgets (Johnson, 1992). Ittner andLarcker (1998) reported more extensive use ofnon-nanc
15、ial performance measures to supple-ment traditional accounting-based measures. Acase study of a UK chemical company imple-menting world-class manufacturing practicesfound that non-nancial measures such as quality,on-time delivery, inventory levels, and productiv-ity replaced the previous emphasis on
16、 budgets andnancial measures (Jazayeri Snell Golhar, Stamm,Johnston, 1989; Plenert, 1990;Schonberger, 1982; Snell Ittner Kaplan Najarian, 1993; Perera, Harrison, Flynn et al., 1995; Mehra Moshavi, 1990; Spencer Young,1992). Moshavi (1990) suggests ve essential ele-ments of JIT: setup time reduction,
17、 focus owprocessing, containerization (pull system contain-ers for inventory), parts control (kanban), andpreventive maintenance. Young (1992) discussesthe JIT manufacturing system, kaizen, total qual-ity control, and JIT purchasing as importantunderlying factors of the Japanese manufacturingsystem.
18、 A literature review through 1990 by Whiteand Ruch found 16 techniques identied as JIT. Aconsensus for10ofthese JITelementswas iteratedby established JIT authors (e.g. Hall, Hay, Mon-den, Schonberger, Shingo, and Suzaki). Theseconsensus elements are described in previousresearch as encompassing JIT
19、practices and areused by White et al. (1999; White Sim Swanson Young et al., 1988).The third JIT factor identied is one of uniquelyJIT practices that describe the extent to whichcompanies have implemented JIT purchasing andkanban. The likelihood is low that companies whoare not fully committed to a
20、JIT program wouldadopt these practices. A description of the specicsurvey questions that support these factors isfound in Appendix B. For results of the factoranalysis for JIT elements, refer to Table 3.3.4. IndependentvariablesNine constructs were selected to examine thenon-traditional performance
21、measures and incen-tive systems of manufacturing rms. Four of theseconstructs, which represent performance measuresfor evaluating manufacturing productivity,measure hypothesis 1A: bottom-up data gatheringtechniques; benchmarking for products, services,and processes; frequency of measurements andrepo
22、rtsonquality; andmanufacturingperformancemeasures. The rst three of these constructs weredened in Ittner and Larckers 1995 TQM study.The last construct is similar to one examined byDurden et al. (1999) in examining the use of non-nancial manufacturing performance indicators ina JIT environment. Thre
23、e constructs related toTable 3Factor analysis (VARIMAX rotation) and factor loadings for JIT variablesaFactor 1JITMANUFFactor 2JITQLTYFactor 3JITUNIQUECronbachs alpha 0.831 0.946 0.684Focused factory 0.740Group technology 0.770Reduced setup times 0.706Productive maintenance 0.668Multi-function emplo
24、yees 0.501Uniform work load 0.731Product quality improvement 0.917Process quality improvement 0.902Kanbansystem 0.820JIT purchasing 0.825aAll loadings in excess of 0.300 are shown.n=253.8All of the 11 elements loaded greater than 0.50 onto one ofthe three constructs except for number 11, asking abou
25、t the useof quality circles. It was evident from initial observations ofthe survey responses that only a few rms (both JIT and non-JIT) used quality circles. Thus, this question was eliminatedfrom further testing representing JIT.R.R.Fullerton,C.S.McWatters/Accounting,OrganizationsandSociety27(2002)
26、711735 719performance incentives through compensation areexamined in hypothesis 1B: compensation ties tonon-nancial performance; compensation ties toquality and teamperformance ; and compensationties to traditional protability measures. The lasttwo constructs of the research analysis testinghypothes
27、is 1C are: communicationofthestrategicplantomiddlemanagers, rst-linesupervisors, andnon-management personnel; and empowerment ofemployeesindecisionmaking.3.4.1. FactorsforperformancemeasuresandincentivesystemsThirty-nine items from the survey instrumentwere evaluated to measure the nine performance-
28、measure and incentive-system constructs. Toreduce and summarize the collected data, thesesurvey items were subjected to a factor analysis.Using the principal components method, the fac-tor analysis revealed ten distinct factors witheigenvalues greater than 1.0, which accounted for73% of the total va
29、riance in the data.9The VAR-IMAX rotation resulted in the following factors:QLTYREV: The frequency with which qualityissues are measured and reportedto management strata.COMPQLTY: The importance of quality andteamwork in determiningcompensation.BOTTOM: The use of bottom-up datagathering techniques s
30、uch asPareto analysis, histograms,and cause-and-eect diagrams toevaluate operations.COMPBDGT: The importance of adherence tobudget items in determiningcompensation.BENCH: The use of benchmarking toevaluate operations.PERFWASTE:The use of performance measuresrelated to waste and ineciencyin evaluatin
31、g the manufacturingsystemSTRPLAN: The extent to which employeesunderstand the rms strategicplan.PERFVEND: The use of performance measuresrelated to timeliness and vendorperformance in evaluating themanufacturing system.COMPNF: The use of non-nancial measuresto determine compensation.EMPOWER: The ext
32、ent to which line managersand non-management personnelare empowered to make decisions.A descriptionofthespecicsurveyquestionsthatsupport these factors is found in Appendix A. Fortheresultsfromthefactoranalysis,refertoTable4.3.5. ControlvariablesFour control variables (covariates) are includedin the
33、regression testing. Firmsize (SIZE) aectsmost aspects of a rms strategy and success;therefore a rms net sales are used to control forrm size. The net sales for each sample rm areobtained from COMPUSTAT data. Whether arm follows a more innovative strategy can aectits willingness to make changes. Inno
34、vative rmsare more risky and generally more protable(Capon, Farle, therefore, it waseliminated from the nal analyses. The expected single con-struct measuring manufacturing performance loaded onto twofactors as per the factor analysis: PERFWASTE andPERFVEND.720 R.R.Fullerton,C.S.McWatters/Accounting
35、,OrganizationsandSociety27(2002)711735Table 4Factor analysis (VARIMAX rotation) and factor loadings for performance measures and incentive systems variablesaFactor 1QLTYREVFactor 2COMPQLTYFactor 3BOTTOMFactor 4COMPBDGTFactor 5BENCHFactor 6PERFWASTEFactor 7STRPLANFactor 8PERFVENDFactor 9COMPNFFactor
36、10EMPOWERCronbachs alpha 0.920 0.909 0.873 0.862 0.907 0.775 0.828 0.783 0.869 0.809TM reviews quality results 0.719TM reviews quality consequences 0.793MM reviews quality results 0.836MM reviews quality consequences 0.857LS reviews quality results 0.819LS reviews quality consequences 0.828MM compen
37、sationquality 0.773MM compensationthroughput 0.703MM compensationteamwork 0.734LS compensationquality 0.854LS compensationthroughput 0.827LS compensationteamwork 0.783Use cause-and-eect diagrams 0.729Use histograms 0.791Use owcharting 0.658Use Pareto analysis 0.778Use scatter diagrams 0.748Use SPC c
38、harts 0.652MM compensationvariances 0.788MM compensationbudget 0.704LS compensationvariances 0.794LS compensationbudget 0.780Benchmarking of operations 0.844Benchmarking of products 0.867Benchmarking of delivery systems 0.836Performance measuresdowntime 0.691Performance measuresscrap 0.721Performanc
39、e measuresrework 0.689Performance measuressetups 0.685MM understand strategic plan 0.791LS understand strategic plan 0.793NM understand strategic plan 0.776Performance measureson-time 0.514Vendor performancequality 0.866Vendor performanceon-time 0.879MM compensationnon-nancial 0.867LS compensationno
40、n-nancial 0.873LS empowerment 0.713NM empowerment 0.741aAll loadings in excess of 0.300 are shown.n=253. TM=top management; MM=middle management; LS=line supervisors; NM=non-management.R.R.Fullerton,C.S.McWatters/Accounting,OrganizationsandSociety27(2002)711735721make major operational changes. If a
41、 rm ishighly centralized, the employees will be much lessinvolved in decision making and organizationalchanges than if it is more decentralized. Kalaga-nam and Lindsay (1998) demonstrated how adapt-ing more organic (decentralized) organizationalstructures led to greater benets from JIT adop-tion. Th
42、e organizationalstructure (STRUCTR) ofa rm is identied on the questionnaire.3.5.1. FactorsforcontrolvariablesThe six survey questions related to rm innova-tion and organizational structure were reducedand summarized using factor analysis. These sixvariables converged into two anticipated distinctfac
43、tors with eigenvalues in excess of 1.0, account-ing for 66% of the total variance in the data. TheVARIMAX rotation resulted in the followingcontrol variables:STRUCTR: The extent of centralization ordecentralization of a rmsorganizational structure.INNOV: Theextenttowhichthermconsidersitself a leader
44、 in product and processdesign and product technology.A detailed description of the specic questionsthat support these control variables is found inAppendix B. Refer to Table 5 for the rotated fac-tor solution.3.6. ConstructvalidityandreliabilityanalysisThe factor solutions for the dened constructssu
45、pport the construct validity of the survey instru-ment. Convergent validity is demonstrated by eachfactor having multiple-question loadings in excessof 0.5. In addition, discriminant validity is sup-ported, since none of the questions in the factoranalyses have loadings in excess of 0.3 on morethan
46、one factor.10Cronbachs alpha is used as thecoecient of reliability for testing the internalconsistency of the constructs validated by the fac-tor analysis. The alpha coecients for all of theconstructs are in excess of 0.7.11(The alpha coef-cients are included in Tables 35.) Overall, thesetests suppo
47、rt the validity of the measures repre-senting the constructs used in this study.4. Research results4.1. DescriptivestatisticsOne objective of this study is to capture thedegree to which the sample rms have imple-mented JIT practices. On the survey instrument,the respondents were asked to provide the
48、 degreeto which they were using 10 individual aspects ofJIT (scaled from 1 to 6). Respondents also wereasked to indicate whether their rm had formallyimplemented JIT. Descriptive statistics depictingthe means for each of the individual elements,along with the three JIT factors, and the totalcombinat
49、ion of the JIT elements are shown onTable 6. The data are presented in terms of thetotal sample, the JIT sample rms, and the non-JIT sample rms.The ANOVA comparison of the means betweenthe JIT and non-JIT rms, found on Table 6,consistently shows highly signicant dierences(p0.000) between JIT and non-JIT rms. Foreach JIT measure, the mean for the JIT rmsexceeds 3.6, whereas only three of the individualelements of the non-JIT rms have a mean greaterthan 3.1. JIT rms have
