27Reporting Manufacturing Porformance Measures

7/29/2019 27Reporting Manufacturing Porformance Measures

1/24

JMARVolume FiveFall 1993

Reporting ManufacturingPerform ance M easures to W orkers: AnEmpirical Study

Rajiv D. BankerGordon PotterandRoger G. SchroederUniversity of Minnesota iAbstract: Much management accounting research has focused on the provi-sion of periodic, aggregated financiai information to managers for planning andcontrol. Recently, many firms have adopted just-in-time production, total quaiitymanagement, and teamwork practices for their manufacturing operations. Thesenew manufacturing practices rely on increased worker involvement in the con-trol of ail phases of m anufacturing, with the expectation that such involvementwill result in the identification of opportunities for process innovations and man u-facturing performance improvements.The central theme of this paper is that the adoption of the new manufactur-ing practices necessitates changes in performance reporting and control sys-tems. Successful implementation of these practices requires the workers to iden-tify w ays to improve the manufacturing proc ess, reduce defects and ensure thatthe manufacturing operations run efficiently. Reporting manufacturing perfor-mance information provides line personnel with the feedback that is necessaryfor learning and directs their efforts to productivity and quality improvements.Demand for shop floor performance reporting systems Is therefore likely to begreater where these new m anufacturing practices are employe d.

Using a sample of 362 worker responses from 40 plants, we docum ent thatthe reporting of manufacturing performance measures to line personnel is posi-tively related to the implementation of just-in-time, teamwork, and total qualitymanagement practices. Worker morale is also found to be positively related tothese new manufacturing practices and to the reporting of performance infor-mation. As s uch, the resuKs provide evidence on the linkage between manufac-turing practices and performance reporting systems.

I. INTRODUCTIONFor most of this centuiy management accounting has focused prima-rily on the provision of periodic, aggregated financial Information to m an-agers for purposes of planning and control. This activity has been influ-enced considerably by external reporting requirem ents [Johnson and Kaplan1987]. During the past decade, however, many firms have made substan-tial adjus tme nts to their app roach es to ma nufacturing tha t necessitate sig-nificant changes ln their control systems. Milgrom and Roberts [1990], forInstance, discuss how many plants are making a number of coordinated


2/24

34 Journal of Managem ent Accounting Research, Fall 1993workforce management policies that shift the control of production to theline personnel on the factory floor. Production problems are not identifiedby m ana gers alone, nor are solutions dictated to workers; instead workersare relied upon to come up with ways to improve process quality and pro-ductivity. Consequently, manufacturing performance information needs tobe reported to shop floor personnel. In this paper, we empirically investi-gate the link between manufacturing practice choices and the reporting ofmanufacturing performance measures to shop Door personnel.

M anufacturing practices of just-in -time prod uction (JFT) and total qualitymanagement (TQM) have recently been adopted by many Japanese andU.S. m anu facture rs. M oreover, man y plan ts now emphasize teamw ork a ndencourage their workers to solve problems and to generate innovative ap-proaches to improve production. Studies of automobile components andother industries described In Voss [1987] emphasize the contrast in orga-nizational practices. Traditional manufacturing plants tend to be laid outby machine or process function. Line personnel, separated from their co-workers by inventory, become specialized by repeatedly processing largebatches of similar materials. Inventories are pushed through the systemwith quality inspections conducted by quality control personnel occurringat the end of production. In contrast, the new manufacturing practicesemphasize products and customers rather than mass production. Prod-ucts are grouped into cells, and workers are assigned to these cells witheach employee performing several functions. Products are pulled throughthe system on demand. Batches are small, there is little work-in-processinventory and tea m s are responsible for prod uction and quality. These newworkforce ma nag em ent practices str ess wo rker involvement [Zipkin 19911.W orkers learn thro ug h doing which re su lts in improved ability to recognizeproblems and to implement solutions LAokl 19861. Reporting productivityand quality information to line personnel p rovides them with the feedbacknecessary for learning and production improvement.

A review of the recent practitioners' literature indicates that manage-ment accountants are becoming increasingly interested in the implemen-tation of new manufacturing performance measurement systems. ' Someaccounting practitioners are beginning to recognize that they must expandtheir horizons and become fully cognizant of the changes that are occur-ring in manufacturing if they wish to retain their position as the primarysource of performance reports in organizations (Howell and Soucy 1987;Simon 1990]. The Increa sing a tten tion to the need for the Involvement ofcontrollers in the reporting of operations-based performance m ea su res rep -

'Thls echoes the call by Kaplan [1983, 1984), and Is reflected also in Lammert and Ehrsam119871 and McNair and MosconI [1987]. Kaplan [1984. p. 414] sta tes :Management accountants may feel that their own area of comparative advantage


3/24

Banker, Potter, and Schroeder 35res ents a retu rn to the origin of m anagem ent accou nting system s, ra the rthan being a new task for the accountant (Johnson and Kaplan 1987].Little is know n ab out the factors influencing a firm's choice of a m an u-facturing performance reporting system [Karmarkar et al. 1987]. In thispaper, we argue th at this choice is tnfiuenced principally by the firm s m an u-facturing practices. Implem entation of TQM, JIT, and teamwork practicesp ut s the control of produ ction ln the h an ds of the workers, an d thereforeincrea ses the value of reporting manu facturing performance informationto line personnel.^ We test th is hypothesis us ing questiormaire resp on sesfrom 362 w orkers in 40 plan ts located ln the U nited S tates.The paper is organized as follows. In the next section we describe thenew manufacturing practices and their Implications for the role of manu-facturing performance information. This Is followed by a discussion of therese arch m eth ods In Section III and em pirical re su lts ln Section IV. Weconclude in Section V with a discussion of the resu lts and some su gges-tions for future research.

U. BACKGROUND AND HYPOTHESESNew Manufacturing Practices

In the past two decades many basic industries in the United Stateshave faced increasing competition from foreign m anufa ctu rers . In resp onse,many U.S. plants have begun to adopt new manufacturing practices toIncrease their global competitiveness. Three manufacturing practices thatare Increasingly being adopted by U.S. plants are total quality manage-m en t (TQM), jus t-in -tlm e ma nufa cturing (JIT) and teamw ork.Total quality management, TQM, promotes involvement of the entireorganization in continuously improving quality. Aspects of TQM includesimplicity of product design, interaction with suppliers, and continuousupkeep of production equipment [Johnson and Kaplan 1987]. More im-portantly, the responsibility for detecting nonconforming items shifts froma quality control d epartm en t to line perso nne l. Young et al. [1988] rfer tothis a s process-ba sed quality control, an approa ch in which quality is builtinto a product by workers as it moves through the factory. TQM makeseach worker responsible for quality control and for stopping productionwhen there is a manufacturing problem [Monden 19891. Workers are en-couraged to identify ways to improve product and process quality. Theirobservations on the shop floor also help fine-tune new product and pro-cess designs [Cole 19831.Closely related to the emphasis on quality is JIT production practice.Jo hn so n and Kaplan [1987]. Maskell [1989] and Schonberger [1986] sta te^Fosterand Homgren [1987, p. 25] state:


4/24

36 Journal of Managem ent Accounting Research, F all 1993

that manufacturing actions consistent with JIT Include running produc-tion on a dem and-p ull basis , placing production control with workers, a ndstreamlining the prod uction pro cess. Young et al. (1988] asse rt th at a keyaspect of the Just- in-tim e/pu ll system is the tight relationship betw eenproduction goals and current production. Implementation of JIT produc-tion is also associated with a reduction in the nu m be r of labor grad es a ndthe employment of multifunction workers.^ As the holding of inventorybuffers declines, grea ter coordination Is requ ired [March and S imon 1958].Responsibility is therefore placed with workers and work teams to controlthe process, resulting In a greater demand for workers to learn multipleJob skills [Monden 1989; Schonberger 1986; Foster and Horngren 1987].The Association for Manufacturing Excellence has published severalreports on another significant change in managing manufacturing opera-tions. Many U.S. pla nts now encourage workers to work In team s to tackleproblems on the shop fioor. Workers are encouraged to pool their knowl-edge of the prod uction proc ess and come up with innovative app roac hes toimprove productivity and quality and to reduce production lead time. Re-ports by Burghard [19901, Puckett and Pacheco [1990], Rhea [19871 andRosen [1989] suggest that plants encouraging small group problem solv-ing on the shop floor have sub stan tially Improved quality and productivity,and significantly reduced defect rates and cycle times.Just-in-time, quality and teamwork are closely related improvement ef-forts that are sometimes classified together as world class manufacturingpractices [Schonber ger 1986; Milgrom and Roberts 1990]. A high level of qualityIs a prerequisite for JIT improvement effori^s, otherwise production will bestopped constantly due to defective p ar ts [Monden 1989]. Young et al. [1988]find tha t the potential performance benefits of just- ln-tlm e are only realizedwhen a quality program is in place. On the other han d. JTT serves to improvequality as Inventory levels are reduced and problems, which were previouslyhidden by inventory, are uncovered. Teamwork is essential in implementingboth JIT and quality practices, as ha s been documented by Schonbei:ger [19861.Imal [1986] provides numerous examples of production teams contributingto quality Improvements at Japanese plants.

The common theme of these new ma nufacturing approac hes is the at-tempt to fully utilize the talents of workers on the shop floor by puttingproduction un der their control. Workers are encouraged to solve problem sand improvise. In a contingency theoretic framework these ap proach es rep-resent a shift from the traditional "mechanistic" organization to an evolv-ing "organic" organization [Lawrence and Lorsch 19671. Workers are nolonger assigned to highly programmed tasks with supervision from above,rath er the y are encouraged to be m ore flexible and interactive. Respect forthe worker is also cultivated under these types of production practicesIMonden 19891.* Zipkin (1991, p. 46] writes;^Foster and Horngren [1987] report that one manufacturer reduced the number of labor


5/24

Banker, Potter, and Schroeder 37

One of the central arguments for JIT has been that it boosts workers'morale while enlisting their efforts In the productivity Improvement pro-cess. JIT has long been Identified as part of the movement towards Em-ployee involvement."Hypotheses

We have argued that JIT. TQM and teamwork practices shift produc-tion control to personnel on the shop floor. Worker control is expected toIncrease worker morale and motivate them to contribute to process im-provements and innovations. One ofthe primary perceived benefits ofthenew manufacturing practices Is the gain in efficiency that occurs from theus e of on-the -spot knowledge an d problem solving abilities of workers [Aoki19861. In order for workers to identify problems and opportunities, andcoordinate their efforts, management needs to provide them feedback in-formation in the form of manufacturing performance measures.^In Juran's Qualitij Control Handbook, Baker [1988. p. 10.28-291 statesthat shop floor workers in the new production systems must be providedthe means for self-control: workers must have the ability to regulate theirwork (self-control), they must know what they are doing (feedback for learn-ing), and they must know what they are supposed to do (goal directinginformation). Performance feedback to workers is nec essa ry to enable them"to determine the relationship between their own behavior and the out-comes the process is producing." Self-control with feedback of results andknowledge of goals makes it possible for workers to discover means to im-prove the production process. 'The freedom afforded individuals causestheir creativity, intelligence, and skill to be challenged."The necessity of feedback for learning is one of the most dependableand most studied findings in the cognitive sciences (e.g.. Manis 1968).Tversky an d K ahne man [19861 assert th at effective learning ta ke s placeonly under certain conditions, and requires accurate and immediate feed-back about the relation between the situational conditions and the appro-priate respon se.^ In addition, records help bolster mem ory, which unaid edmay be subject to a number of selectivity biases (Hogarth 19801.

The value of shop floor information is also consistent with organiza-tional behavior research that has shown that feedback helps promote taskoriented behavior [Ashford and Cummlngs 1984; Ilgen et al. 1979]. Ap-pealing to prior goal theory r ese arch stu die s. Wexley and Yukl [19841 rec-ommend that employees should have specific performance goals to guideliterature on total quality management IDemlng 1986: Ebrahlmpour and Lee 1988;Garvin 1 983 . 1986; Sehonb erger 1986; Tagucbi and Clausing 1990; Take uchi and Q uelch19831 also sugg ests tha t firms with successful quality program s provide specific Informationto workers on process outpu t. Garvin [1983] states: "Successful monitoringof quality assu mesthat the necessary data are available, which Is not always true. Without speclfie and timely

Information on defects and field failures, improvements In quality are seldom possible."Schonberger [1986. p. 32] states: There Is substantial benefit from (shop noorinrormatlon)...Informationofvalue to the company does not stay in people's h eads . It eomes


6/24

38 Journal of Managem ent Accounting ResearcK Fall 1993behavior. Jo hn so n an d K aplan [1987]. Peters and Waterm an [1982. p.267]and Utzlg [1988] also suggest that Information can be used to promotedesired employee behavior. Govindarajan and Gu pta [1985] state t ha t whenperceived rewards are attached to specific performance measures, behav-ior is guided by the desire to optimize those performance m ea su re s. There-fore, shop floor performance information may be used to promote specificworker behavior [Daniel and Reitsperger 19911. Since the new manufac-turing practices rely on the workers for process improvements, their ef-forts may be guided in this direction by providing them quality and pro-ductivity information.We hypothesize therefore that the perceived value, and hence the pro-vision of m anufactu ring perform ance Information to line personnel, is posi-tively related to just-in-time, quality, and teamwork practices.H J: The availability of information on productivity and quality is positivelyrelated to the extent of implem entation of just-in- tim e, quality, andteamwork programs.

The above hypothesis addresses Kaplan's suggestion that firms shouldprovide information on productivity and quality. It is silent, however, onthe form and content of such information. Quality and productivity infor-mation may include customer satisfaction surveys, field failure rates, valueadded per employee and other plant-wide efficiency m ea su res. On the othe rhand, charts displaying defect rates, schedule compliance and machinebreakdowns represent information that can be identified easily with spe-cific production cells or work stations.

Analytic tools for process control, such as process flow charts. Paretoanalysis plots, fishbone charts, histograms, run diagrams, control chartsand scatter diagrams also require the posting of specific performance in-formation on the shop floor [Ishikawa 1972; Schonberger 1986. Chapter7]. As workers acquire greater responsibility in the new manufacturingenvironm ent to control their production sch edules and to inspect their ownoutp ut, posting of such ch arts becom es importan t for providing them withthe necessary feedback information. Therefore, we also test a detailed hy-pothesis concerning the provision of visual chart information about spe-cific sho p floor operation s.H gi The posting of ch ar ts abo ut defects, sched ule compliance and m ach inebreak dow n on the sho p floor is positively related to the extent of Imple-m entatio n of just-in -tim e, quality and team work prog ram s.

Finally, we note that these new manufacturing practices rely criticallyon worker involvement. Klein [1989] and Zipkin [1991] no te an increase inworker s tre ss and decline in morale in some pla nts. Womack et al. [1990].however, find worker morale to be higher in auto assembly plants thathave adopted the new practices. Clearly a drop In worker morale wouldargue against increases in economic efficiencies stemming from worker


7/24

Banker, Potter, and Schroeder 39m . RESEARCH METHOD

Sample PlantsThe data used in this researc h consist of question naire re spo nse s from362 workers from 40 man ufacturing plan ts situated in the United States.^To co nstr uct this data ba se Initially 60 p lan ts were randomly selected fromlists of manufacturing plants stratified to equally represent the transpor-tation equipment, electronics and machinery industries. The plant manag-ers were contacted by letter and telephone to solicit participation in thestudy. For the 42 plants that agreed to participate, questionnaires wereadministered to ten randomly selected workers at each plant. Two plantsretumed fewer than five worker surveys and are excluded from the analy-sis. The actual number of worker surveys retumed from the 40 plants

included in this stud y ranged from 6 to 10 per plant. The que stionn aireswere collected directly to en su re confidentiality of worker res po nses . Sum -mary information on costs, employment, square footage and capacity utili-zation for the 40 pla nts is presen ted in Table 1.Twelve of the plants were visited by the research team in the initialphase of this study. These site visits included a tour of the facility andinterviews with the plant manager, quality manager, plant accountant, aproduc tion manager, a process engineer, supervisors an d workers. The ques-tionnaire instruments were also pretested during these site visits. In manyof the plants that were visited, charts and graphs were displayed on theshop floor. For example. In one plant each dep artm ent had bulletin bo ard swhich showed the latest production statistics. These Included several dif-ferent statistical process control and quality charts reporting defect ratesfor the parts produced by that work center. Pareto charts, cause-and-ef-fect diagrams and run charts for certain critical parts [e.g., Deming 1986;Ishakaw a 1972]. Ch arts displaying daily sched ule com pliance (planned vs.actual schedule), ch art s on ma chine breakdow ns and m ainten ance actlvl-^Slnce our research focus is on manufacturing performance Information provided to workers.our tests are based on worker responses. While workers were the principal informants,secondary information was collected from supervisors and managerial staff to validate workerresponses.

Table 1Descriptive Information About Sample Plants(n=40)Quartiles

VariableTotal Plant EmploymentTotal Manufacturing Costs(millions)

Mean6 3 8

$74.8

StandardDeviation7 7 9

$81.12 5 %1 8 2

$24.06 0 %4 1 9

$41 . i7 5 %7 1 5

$104.6


8/24

40 Journal of Managerrvent Accounting Research. Fall 1993

ties, and charts pertaining to weekly productivity measures were also evi-dent. In one plant, quality control and production Information was dis-played in realtime on a computer screen. It was clear that a great deal ofvisual Information was being provided to the workers on the shop floorThe plant visits also revealed that the extent of implementation of thenew manufacturing practices varied within the plant. Most plants beganthe implementation for specific production cells and product groups, andgradually extended the practices to the rest of the plant. There were alsodifferences between different production departments within plants thatwere not organized into cells. For example, in one plant we found the work-ers In the fabrication department using elaborate defects and productioncharts, while the workers In the final assembly department were Just be-ginning to display these charts. The workers in the fabrication departmenthad completed their training courses in statistical process control meth-ods, while workers in other departments were yet to take these same courses.Because of such variations in the implementation of the new manufactur-ing practices within plants, we conduct our analysis at the level of indi-vidual worker responses.Independent Variables

The independent vartables in our model are the following four scalemeasures constructed to represent JIT, TQM. and teamwork practices, andthe extent of decentralization:1) Jrr = A scale measuring use of JIT manufacturing practices.2) TQM = A scale measuring Implementation oftotal quality program.3) TEAMWK = A scale measurtng use of small teams for problem solvingon the shop floor.4) DECENT = A scale measurtng decentralization of authority.The questions Included In each scale are presented In Table 2. The actualquestionnaire included these questions in a randomly assigned order. Eachquestion within a scale Is measured on a five point Likert scale, whereeach worker was required to respond on the range "strongly agree = 5" to"strongly disagree = 1." Scales are constructed by equally weighting thequestions.The first scale measures key aspects of JIT implementation from aworkers perspective: compliance to a dally production schedule and learningmultiple skills. Young et al. [1988] state that a just-in-time practice main-tains a tight relationship between current production and production sched-ule goals. Schonberger 119861 asserts that employee involvement In pro-duction and setups requires them to learn multiple skills. The quality scale(TQM) measures key aspects of a quality program: quality incentives, workerinspection of output, and stoppage of production for quality problems. Younget al. [19881 argue that key aspects of a quality program require that all


9/24


Table 2Scale Measures* for Manuiacturintf Practices1) J I T S ca l e

I) O u r s c h e d u l e Is d e s i g n e d to a l low t ime fo r c a t c h i n g u p . d u e t o p r o d u c t i o ns t o p p a g e s fo r q u a l i t y p r o b l e m s .li) D i rec t l ab or un de rgo es t r a in in g to pe r f o r m m u l t i p l e t a sk s i n t he p r o d u c -t i on p r oce s s .111) Plant employees a r e r e w a r d e d for l e a r n i n g n ew sk i l l s .iv) W e u s u a l l y m e e t t h e p r o d u c t i o n s c h e d u l e e a c h d a y .2) TQM S c a l eI ) Workers a r e r e w a r d e d fo r q u a l i t y i m p r o v e m e n t .II) If I i m p r o v e q u a l i t y , m a n a g e m e n t wiU r e w a r d m e .i i i ) P roduct ion i s s toppe d I m m e d i a t e l y fo r q u a l i t y p r o b l e m s .Iv) I i n s p e c t m y o w n o u t p u t . "3 ) TEAMWK Sc a lei) D ur i ng p r ob l e m so l v ing s e s s i on s , w e m a k e a n effort to ge t a l l t e a m m e m -b e r s ' o p i n i o n s a n d i deas be fore making a d e c i s i on .ii) O u r p l a n t f o r m s t e a m s to so l v e p r ob l e m s .Iii) In the pa s t t h r e e y e a r s , m a ny p r ob l e m s h a v e b e e n sol v ed t h r ou g h sm a l lg r o u p s e s s i o n s .4 ) DE CE NT S ca l ei) I c a n d o a l m o s t a n y t h i n g I w a n t w i t h o u t c o n s u l t i n g m y b o s s .ii) Ev en sm a l l m a t t e r s hav e t o be referred to s o m e o n e h i g h e r up for a final

answer (Rever se sca le ) .i ii ) Th is pl an t i s a good place for a p e r s o n w h o l ikes to m a k e h i s o w n d e c i -s i o n s .tv) A n y d e c i s i on I m a k e h a s to h a v e m y boss ' s approva l (Rever se sca le ) .v ) There c a n b e l i t t le act ion taken here unt i l a supe r v i so r a pp r ov e s a d e c i s i on(Reverse sca le) .'Each response is measured on a scale from "strongly agree = 5" to "strongly d is-agree = 1.""Based on factor analysis this question Is omitted from the scale.their supervisors. This scale was developed and validated by Alken andHage 119661 to m eas ure h ierarchy of authority. A uthorizing w orkers to m akeproduction decisions is an important aspect of these new manufacturingpract ices.We examined the four scales for reliability (consistency) and validity.Reliability addresses the extent to which measures or responses are de-pendable, or free of error [Nunnally 1967: Kerlinger 19731. Scale reliabilityis evaluated by examining C ronbac h's alph a. This coefficient is ba sed onthe correlations am ong the resp ons es com prising a scale. The alpha coeffi-cien ts for the m anufacturing practice scales are presented in Panel A ofTable 3. They range from a low of 63.5 percent for JIT to a high of 77.2


10/24

42 Journal of Management Accounting Research, Fall 1993

Table 3RelUbUity and Validity Testsof Manufacturing Practice ScalesPanel A. Reliability and Construct Validity SUtisUcs

Cronbach's AlphaProportion of Variance

Loading on largest FactorNo. of Factors withEigenvalue > 1

JIT.635.4761

TQM.772.6321

TEAMWK.701

.611

1

DECENT.739.4791

Panel B: Consistency StatisticsVariation in Worker ResponsesExplained by Plant EffectSpearman (and Pearson)Correlation Between AverageWorker and Supervisor/Manager ResponsesPearson correlations are In parentheses.

JIT.275**

.603**

.550)**

TQM.486**

.743**(.763)**

TEAMWK.338**

. 709"(.697)**

DECENT.296**

.478**(.616)**

Panel C: Pearson Partial Correlations (After Industry Control) for Criterion-Related Validity

JIT TQM TEAMWK DECENTCycle Time -.336* -.183 -.223Supplier Certification .418** .356* .157Rework -.281 - . 351* .131

-.125.472**.008

Cycle timeSupplierCertificationRework

Logarithm of number of days from receipt of raw materials untilcustomer receipt of product.Percent of suppliers that are certified.Percent of products requiring rework at final Inspection.

Panel D: Spearman (and Pearson) Correlation with Responses of Supervisors.Plant Managers, and Production Personnel

JIT ProductionKanban r*uU System

JIT-.380**(.388)**.251

TQM.322*(.307)*.243

TEAMWK.549**

(.449)**.171

DECENT.088(.160).005


11/24


Table 3 (Continued)JIT TQM TEAMWK DECENT

Employee Involvement .454** .411** .660** .289*(.409)** (.361)* (.610)** (.431)**Pearson correlations are in parentheses.*Indlcates significant at 5% level (one-tail).*Indicates significant at 1% level (one-tail).

JIT Production = Response of supervisors, process engineer and plant produc-tion manager to question "all major department heads withinthe plant work towards encouraging just-in-tlme production."Kanban Pull = Response of supervisors, process engineer and plant produc-System tion manager to the question "we use a Kanban pull system forproduction control."Authorization = Response of supervisors, process engineer and plant produc-for Quality tion manager to the question "direct labor is authorized to stopProblems production for quality problems."Worker Control = Response of plant production manager to the statement "pro-duction control is ln the hands of workers."Employee = Response of supervisors, process engineer and plant produc-Involvement tion manager to the question "our top management stronglyencourage employee involvement ln the production process."Each question is measured on a five-point Likert scale where "strongly agree = 5"and "strongly disagree = 1."

While the responses of workers at the same plant may differ due tovarying involvement in key aspects of the manufacturing practices imple-mented in different parts of the plant, there is likely to be some consis-tency between the responses from the same plant. The percentage of varia-tion in worker responses explained by the plant effect ranged from 27.5percen t for JIT to 48.6 p ercen t for TQM, the plant effect was highly signifi-cant (p < .001) for all four scales. We also compared the average workerresponse at each pla nt with the average of respo nses of tw o/th ree supervi-sors from the same plant. The results presented in Panel B of Table 3indicate that all the correlations are highly significant (p < .001). rangingbetween .478 for DECEMT and .743 for TQM.We evalua ted the practice scale s for conten t validity, co nstruc t validity,and criterion-related validity [Kerlinger 1973]. Content validity is strivedfor by constructing questions that represent key aspects of the new m an u-facturing practices. We have discussed earlier why the scales representessential attributes of the new manufacturing practices from a worker'sperspective. Construct validity is supported by factor analysis. Panel A ofTable 3 reports that each scale loads on one factor with the percentage ofthe v ariation explained by that factor ranging from 47.6 percent for JIT to63.2 percent forTQM.^


12/24

44 Joumal of Managem ent Accounting ResearcK Fall 1993ables that are expected to be highly correlated with the practices. Partialcorrelations are used to examine the correlations after controlling for in-dustry representation. The criterion variables are plant-wide measures ofcycle time (ln days), percent of units requiring rework and percent of sup-pliers certified. Cycle time is chosen because a key benefit of JIT Is thereduction of inventories. Reduction of rework requirements is a primarygoal of quality improvement prog ram s and s upp lier certification is expectedto improve Input quality and facilitate JIT production. B ecause thes e thre evariab les are mea sure d at the plant level, the resp ons es of work ers from aparticular plant are averaged for this analysis. These correlations are pre-sen ted in Panel C of Table 3. With the exception of the insignificant posi-tive partial correlation between DECENT and rework, and TEAMWK andrework, all correlations are in the expected direction and five are signifi-can t at the 5 percent level. We also com pare ou r new m anu facturing prac -tice variables with the responses of plant management to workforce re-lated q uestion s in Panel D of Table 3. The practice sc ales are strongly re-lated to managers" perceptions of JIT production, authorization for qualityproblems, worker control and employee involvement. This suggests thatour scale measures based on worker responses are capturing importantattribu tes of the new manufacturing practices in place at p lants.Dependent Variables

Measures of availability of information and level of morale were alsosolicited from workers in the questionnaire surveys. All of these questionsare presented in Table 4. Similar to the independent variables, each ques-tion was measured on a five point Likert scale, where responses rangedfrom "strongly agree = 5" to "strongly disagree = 1." The first two informa-tion availability question s correspond to recommendations by Kaplan (1983]that information on quality and productivity should be compiled and re-ported to employees. The last three represent specific types of charts thatare posted on the shop floor for feedback purposes. Each worker also re-sponded to six questions tha t are compiled into a scale represen ting workermorale. This scale is constructed from a subset of Mowday and Steer's(19791 commitment scale. The scale appears to have reliability and con-struct validity as the alpha coefficient is 83.9 percent and the largest fac-tor explains 56.7 percent ofthe variation.The percentage of variation in worker responses that is explained bythe plant effect ranged from 22.4 percent for machine breakdown charts to34.5 percent for defects charts, all significant at the .001 level. Averageworker responses are consistent with average supervisor/manager re-sponses from the same plant. All correlations are highly significant, andrange from .338 for machine breakdown charts to .729 for defects charts(see Table 5).Estimation Model


13/24


Table 4Descrip tion of D epen dent Variables*Quality and Productivity lT\formation Avaitaldlity1. Qu ality = Information on qua lity perform ance Is readily available to em -ployees.2. Produ ctivity = Information on productivity is readily available to emp loyees.

Posting if Charts on the S hop Floor3 . Defects = Cha rts showing defects are posted on the sho p floor.4. Schedule = C harts showing schedule compliance are posted on the shop

floor.5. Machine = Ch arts plotting the frequency of ma chine breakd ow ns areposted on the shop floor.Worker M orale6. Worker Morale Sc ale:"i) I am willing to put in a great deal of effort beyond tha t norm ally expected .11) I talk u p th is organization to my friends a s a great organization to work for.lil) I would accep t almost a ny type of job a ssignm ent in order to keep workingfor this organization.Iv) I am p rou d to tell oth ers I am part of this organization.v) This organization really Inspires the be st ln me ln the way of Job perfor-mance.vl) I am really glad i chose th is organization to work for over othe rs.

'E^ch variable is measured on a ivepoint Likert scale rang ing from "strongty agree= 5" to "strongly disagree = i.""C ron bac h alpha is .839 an d percent of variation explained by the largest factor is.567.where:Y = quality information, productivity information, defects charts, sched-ule compliance charts, machine breakdown charts, and worker mo-rale, j = 1. ... 6:X = JIT . TQM. TEAMWK. DECENT, i = 1. ... 4;w = worker w. w = 1, ... 362.Two issue s had to be resolved in order to estimate the relation betweenthe manu facturing practices and the depen dent variables. One concern isthe potential correlation among the error terms of workers located in thesame plant. Following Johnston [19841. we specify this relationship as anerrors components model. Specifically, each worker's response error. ^,is assume d to comprise two components: [L , a plant-wide fbced effect term,and 6j^. an independently and identically distributed worker specific ran-dom error term , so tha t j^ = Mjp + V ^^ worker w in plan t p. Conse-quently, dummy variables representing each plant are introduced into themodel to account for the plant effect term:"'


14/24

46 Journal ojManagetnent Accounting ResearcK Fall 1993

8Ia

l O

IsCO

C:J

(N

0)CO

in

C OC N

OoCO

i n COCO OCO to

00 PCO QOO

o ino> o

0) inM CON t

0 ) O Tin in

o inCO in

1 ^ I K .

Si cC5

o o^ c


15/24


39+2^ )j.jpDpw+6)w. (2)

1=1 p=iwhere Dp^ = 1 If worker w is ln plan t p . and = 0 otherwise. Industry differ-ences can be examined by comparing the average plant effect terms byIndustry.A second issue Is that the worker responses to the survey questionsare ordered categorical variables [Maddala 1983]. While the worker moralescale is the average of six respo nses a nd a ppro ach es a con tinuo us variable(takes on 25 possible values), the worker respon ses to the dep end ent vari-ables c oncerning shop floor Information ca n ta ke on only five discrete val-ues. Following Maddala [1983. pp. 46-491 and Cox and Snell [1989. pp.158-161] we also estimate the parameters using a maximum likelihoodnonlinear ordered polychotomous loglt estimation routine.*' The left handside of the estimation model in (2) now represents the proportional oddsfor each ordinal level [McCullagh 19801. These estimates are consistentand asymptotically, normal and efficient. We con struct likelihood ratio tes tsfor overall model fit. and employ a m ea su re of goodness-of-fit. R^ = 1 - [like-lihood ratiop/", suggested by Maddala [1983. p. 39]. ^ i^e results of thisestimation are presented ln Table 9 for comparison with OLS estimatesreported in Table 8.

I V . EMPI RI CAL RE SU LT S .Descriptive StatisticsSum m ary Information on the explanatory variables is presen ted in Table6. Panel A indicates that the means of the scale predictor variables rangefrom a low of 2.91 for TQM and DECENT to a high of 3.42 for TEAMWK.The m ea n JIT respon se is close to that of TQM at 2.97. The quartiles andstandard deviation information indicates that all four scales have a similardegree of variation. Spe arm an and Pearson correlations are reported inPanel B of the table. The correlations document significant positive asso-ciations among the variables representing the new manufacturing prac-tices of JIT. TQM and TEAMWK. This is expected, given tha t Just-in -tim e.quality and teamw ork programs are complementaiy practices that are som e-times grouped together as world class manufacturing practices. The sig-nificant positive association with the decentralization variable is also con-sistent with the view that the new manufacturing practices give line per-sonnel m ore control over the day-to-day operations.Summary data on the dependent variables are presented in Table 7.The Panel A univarlate statistics reveal that average worker resp ons es forthe availability of quality and productivity Information are greater th an theaverage responses for the posting of charts for defects, schedule compli-ance and m achine break dow ns. Having a cha rt on defects or schedule com-


16/24

48 Journal of Managem ent Accounting Research Fall 1993

Descriptive Information on theManufacttirlnjE Practice Scales(n=362)Panel A. Summary Statistics

Variablej r rTQMTEAMWKDECENTPanel B: Correlation MatrixJ ITTQMDECENT

Mean2.972.913.422.91

JIT1.000.509*.422*.141*

StandardDeviation.70.75.8 1.75

TgM.511*1.000.375*.123*

2 6 %2.502.333.002.40

6 0 %3.003.003.673.00

TEAMWK.415*.342*i.OOO. 266*

7 5 %3.503.674.003.40

DECENT.150*.096.240*1.000

Spearman correlations are above the diagonal. Pearson correlations are below thediagonal.**Indicates significant at 5% level.quality or productivity that Is available to the worker. The availability ofmachine breakdown charts is much lower than the availability of defectand schedule compliance charts, suggesting that reporting information onbreakdow ns m ay depend on the m achine m aintenance policy of the plant.Panel B of Table 7 presents pairwise Spearman and Pearson correlationsbetween the de pen den t variables. The significant correlations suggest tha tall of the information variables are driven by similar economic factors.Worker morale is also positively related to each of the workers" informationresponses.Hypotheses Tests

The results of the regressions relating quality and productivity Infor-mation, posting of charts and worker morale to the new manufacturingpractices are presented in Tables 8 and 9. The estimation results with thepolychotomous ordered loglt model are very similar to the OLS results,and therefore we only discuss the OLS results here. All of the regressionsIn Table 8 are significant. The a m oun t of variation ln the worker resp on sesthat can be explained by the predictor variables ranges from 32.8 percentto 52.3 percent.The mo st striking result a cros s all the regression s is the co nsistency of


17/24


N N O CO14 00 ' Q DCO CO CO M

O CO N OCD rr cs ocs CO cs O

CO

o o o oo o o o

CO00

CO CO

o oo o O O COO O COCO CO N cs


18/24

50 Journal of Management Accounting ResearcK Fall 1993

H V

** T2 **

a c *l-H

ooV

CN O CO

in ^CD CNN CO COo -

O O COCO

in -HCO Orf h OCD Vo0 1COo

oqV

C O - , O . - . > - Q 0 - , M


20/24

52 Journal of Management Accounting ResearcK Fall 1993

for the predictor variables do not reverse signs across equations. The sta-tistical significance of the individual coefficients does vary by type ofnonfinancial information.^'* The JIT, TQM and TEAMWK scale-measuresexhibit strong positive association with the likelihood that infonnation onquality and productivity is provided to workers. DECENT is not signifi-can tly related to the provision of this information. All of the p redicto r vari-ables are significantly related to the posting of schedule compliance c har ts,while only JIT and TQM are significantly related to the provision of ma-chine breakdown charts. The regression result for defects charts indicatesthat provision of such information is significantly related to JIT and DE-CENT. Information on defects m ight be expected to be mo re closely relatedto TQM than to the other manufacturing practices. Recall, however, thatthe pairwise correlation between JIT and TQM is 51 percent. On re-estl-mating the defects regression in Table 8 after excluding the JIT variable,the coefficients on TQM and TEAMWK rise to .227 (p=.O22) and .122(p=.O94), respectively. Finally, a F-test on the plant coefficients Indicatesthat in each regression there is also a significant plant effect. A compari-son of average plan t effects, however, does not reveal a significant differ-ence between industries.Table 8 also presents evidence on the strong positive relationship be-tween worker morale and the new manufacturing practices. In fact, over50 percent of the variation in morale is explained by the regression vari-ables. JIT and TEAMWK are particularly strong explanators. It contradictsthe c onc ern s expressed by Zipkin [19911 an d others th at worke rs experi-ence lower morale with the introduction of the new manufacturing prac-tices. This is reassuring as these practices rely on workers to Identify im-provement opportunities when they are given more control over the day-to-day production.

V . C O N C L U DIN G R E M A R K SThis pap er examines the association betw een just-in-time . quality, andteamwork practices and the provision of manufacturing performance in-

formation to line personnel. Because these new organizational and workforcemanagement practices put the control of production in the hands of linepersonnel, we posit a positive relationship between the extent of imple-mentation of these practices and the provision of information to workerson the shop fioor. We present evidence on th is relationship using a sam pleof 362 w orker respo nses from 40 plan ts located in the U nited Sta tes.Footnote 13 (Continued)number of observations d ropped ranged between 12 and 17 for the six regressions. TTie re-estlmatlon Increases the R^ for all the regressions. The R^ now range between 43 percentand 58 percent. All the earlier significant coefilcients remain significant. In addition, two

more coefncients becom e significant at th e 5 percent level: the JIT coefficient In the qualityregression and the TEAMWK coefficient in the machine breakdown regression.


21/24


Two imp ortant findings emerge from this re search . First, we do cum enttha t the provision of information to sho p fioor workers is positively relatedto the im pleme ntation of Just-In-tlme. quality, and teamw ork pra ctices. Assuc h, o ur res ults provide evidence of a link between m anu facturing pra c-tices and control systems that emphasize the role of workers. This is im-po rtant given tha t little is known about factors that influence firms' choice sof performance reporting systems. Second, we establish that employeemorale is positively related to the existence of jus t-in -tim e, qu ality, andteamwork produ ction ^ s t e m s , and the provision of shop floor performanceInformation. T his finding is useful given that there have been con trary opin-ions expressed concerning the impact of these new practices on workerstress and morale (Klein 1989; Zipkin 1991].This study also raises a number of questions for future research. Oneinteresting avenue is to determine if firms that provide performance infor-mation to workers on the shop floor experience more productivity gainsfrom th e implementation of these new m anufa cturing practices tha n firmstha t im plement these pra ctices bu t do not provide shop floor performanc einformation. Another direction for future research is to examine the at-tributes of manufacturing performance reporting systems that promoteworker learning . While it is well known th at feedback is a necessary condi-tion for learning , the im pact of different types and forms of feedback, andthe timing of feedback (on learning), need to be examined in this environ-m ent. Both H ayes an d Clark [1985] and Teece and W inter 119841 believeresearch on learning is critical for advances in modem management. Webelieve research on the link between manufacturtng performance report-ing and learning is important for m ode m mana gemen t accounting. F utureresea rch stud ies may also include detailed d ocum entation of reporting sys-tems that have been adapted to the new manufacturtng practices, and ex-amine transition problems encountered in changing from the old to thenew systems. An important research question is whether the changes inthe shop floor performance reporting system s result from the wo rkers' owninitiative in the new manufactuing environment, or whether the changesare dictated by plant management and corporate accountants. Empiricalresearch isju st beginning in this important area of ma nagem ent accoun t-ing.


22/24

54 Journal of Management Accounting ResearcK Fall 1993REFERENCES

Aiken M., and J. Hage, "Organizational Alienation: A Comparative Analysis." American Socio-logical Review (33, 1966). pp. 497-50 7.Ashford, S. J., and L. L. Cummings, "Feedback as an Individual Resource: Personal Strate-gies of Creating Information," OrgcmizationcdBehciviDr and Hurnan Performance {32. 198pp . 370-398.Aoki. M., "Horizontal vs. Vertical Information St ruc ture of the Firm." The American EconomicReuteu; (December. 1986), pp. 97 1-9 83 .Baker. E. M., Managing Human Performance, Section 10 in Juran's Quality Control Hand-book. J. M. Juran and F. M. Gryna, Eds. (McGraw-Hill Inc.. 1988).Belsley, D. A.. E. Kuh. and R E. Welsch. Regression Diagnostics (John Wllc?y & Sons. 1980).Burghard. F., "Working Together the Winning Way," Target, Journal of Association Jor Marm-factuiing Excellence (Spring. 1990). pp. 28-29.Chapanis, A.. "Knowledge of Performance as an Incentive in Repetitive, Monotonous Tksks,"Journal of App lied Psychology (48, 1964). pp. 263 -267.Cole. R E., "Improving Product Quality Through Continuous Feedback," Management Re-view (October. 1983), pp. 8-17.

Covaleski, M., and M. Aiken, "Accounting and Theories of Organizations: Some PretlminatyE v i d e n c e , " Accounting. Organizations and Society ( 1 1 , 1 9 8 6 ). p p . 2 9 7 - 3 1 9 .Cox, D. R, and E. J. Snell, Ana lysis of Binary Data (Chapman and Hall, 1989).Daniel, S. J., and W. D. Reitsperger, "Linking Quality Strategy with Management ControlSystems: Empirical Evidence from Japanese Industiy," Accounting. Organizations arvlSociety [\9QI).Deming, W. Edw ards, Oulq/" the Crisis (Cambridge, Mass: MITCenter for Advanced Engineer-ing Study, 1986).Drucker, P. F., Th e Em erging Theory of Manufacturing." Harvard Business Review 68 (May-June . 1990). pp. 94 -102.Ebrah lmpo ur. M., and S. M. Lee, "Quality Management Practices of American and Ja pa ne seElectronic Firms In the United States," Production ani Inventory Management Journal(Vol. 29 , No. 4. 1988), pp. 28-31.Eccles, R C , "The Performance Measurement Manifesto," Harvard Business Review (Janu-aiy-February, 1991), pp. 131-137.Einhom, H. J., and R M. Hogarth. "Confidence in Judgment: Persistence of the Illusion ofValidity." Psychological Review (85, 1978), pp. 3 95-416.Foster, G.. and C. T. Homgren, "JIT: Cost Accounting and Cost Management Issues," Man-agement Accounting {June, 1987), pp. 19-25., and , Cost Accounting a Managerial E>nphasis, 7th ed. (Prentice-Hall Inc., 1991)Garvin. D. A.. "Quality on the Line," Harvard Business Review 6\ (September-October, 1983),pp. 65-75., "Quality Problems, Policies and A ttitudes in the United S tates and J ap an : An Explor-atory Study. Academy ofManagenier^ Journal (29, 1986), pp. 65 3-67 3.Goldberg, L. R., "Simple Models or Simple Process? Some Research on Clinical Judgment,"American Psychologist (23, 1968), pp. 48 3-49 6.Govindarajan, V.. and A. L. Gup ta, "Linking Control Systems to B usiness Unit Strategy: Im-pact on Performance," Accounrtng. Organizations and Society (11, 1985), pp. 51-66.Hayes, R H., and K. B. Clark, "ExplainingObserved Productivity Differentials Between Plants:Implications for Ope rations Research,"/n/er/aces (November-December. 1985), pp. 3-14.Hogarth. R Judgment and Choice (John Wiley and Sons, 1980).Howell, R A., and S. R Soucy, 'The NewManufacturing Environment: Major Trends for Man-agement Accounting," Management AccourUing (July, 1987), pp. 21 -27.Ugen, D. R, C. D. Fischer, and M. S. Taylor, "C onsequences of Individual Feedback on [3ehav-ior In Organizations," Journal of Applied Psychology (64, 1979), pp. 349-371.Imai. M., Kaizen. the Key to Japan's Competitive Success (McGraw-Hill Inc.. 1986).Ishfkawa, K.. Guide to Quality Control (Asian Productivity Organization, 1972).Joh nso n. H. J., and R S. Kaplan, Relevance Lost: The Rise and Fall of ManagementAccount-ing (Harvard Business School Press. 1987).Johnston, J., Econometric Methods, 3rd ed. (McGraw-Hill Inc., 1984).Kaplan, R S.. "Measuring Manufacturing Performance: A New Challenge for M anagerial Ac-counting Research," The Accounting Review (October, 1983), pp. 686-705.


23/24

Banker, Potter, and Schroeder 5 5Klein, J. A., "I\\e Human Cost of Manufacturing Refomi," Harvard Business Review (Maruh-April, 1989). pp. 60-66.Lammert, T. B., and R Ehrsam, "The Human Element: The Real Challenge in ModernizingCost Systems," Manc^ement Accounting (July, 1987), pp. 32-37.Lawrence, P. R, and J. W. Lorsch, Organtzatton and Envirorwient: Managirig D^erentiationand IntegraHon (Harvard University Press. 1967).Maddala, G. S., Limited-Dependent Variables and Qualitative VariatAes in Econometrics (Cam-bridge University Press, 1983).Manis, M.. Cognitive Processes (Brooks/Cole Publishing Company, 1968).March, J. G., and H. A. Simon. Organizations (John Wiley & Sons, Inc., 1958).Maskell, B. H., "Performance Measurement for World Class Manufacturing," ManufacturingSystems (July, 1989), pp. 62-64.McCullagh. P., "Regression Models for Ordinal Data," Journal of the Roual Statistical Soceitu(B42 1980). pp. 109-142.McNalr, C. J., and W. Mosconl. "Measuring Performance In an Advanced Manufacturing En-vironment," Management AccountbTg (July, 1987). pp. 28-31.Milgrom, P., and J. Roberts, 'The Economics of Modem Manufacturing: Technology, Strat-egy, and Organization." TheAmerican Economfc Reuieu '(June. 1990), pp. 511 -528.Monden, Y., "Framework of the Just-ln-Tlm e Production System," Jap an es e M anagement Ac-counting: A World Class Approach to ProJU Managetrusnt. Y. Monden and M. Sakurai, Eds.(Productivity Press, 1989).Mowday, R, and R M. Steers, The Measurement of Organizational Commitment." Journal ojVocational Behavior [\A, 1979), pp . 224 -227.Nunnally. J.. Psychonyetric Theory (McGraw-Hill, 1967).Peters, T. L.. and R H. Waterman, In Search of EJccellence: Lessons from America's Best RunCompanies (Harper & Row, 1982).Puckett. J . H.. and J . Pacheco. "Excellence at XEL Communications: Self-directed Work Teams,"Target. Journal oJ A ssociation for Manufacturir^ Excellence (Spring. 1990), pp. 22-27.Rhea. N. W., "People Make Productivity Work ln a Jet Engine Parts Plant," Target Journal oJAssociationJor Manufacturing Excellence {Spring. 1987). p p. 35-3 8.Rosen, N.. Teamwork and the Bottom Line: Groups Make a Differervx (Lawrence Erlbaum

Associates, 1989).Schonberger, R J., World Class Manufacturing: The Lessons oJSimplicity Applied (The FreePress, 1986).Simon. H. A., "Information Technologies and Organizations," The Accounting Review (July,1990), pp. 658-667.Taguchi, G., and D. Clausing, "Robust Quality," Harvard Business Review 68 (January-Feb-ruaiy. 1990). pp. 65-75 .Takeuchi. H., and J. A. Quelch. "Quality is More than Making a Good Product," HarvardBusiness Review [61. 1983), pp. 139-145.Teece, D. J. , and S. G. Winter, 'Th e Limits of Neoclassical Theory In Managem ent Edu eation,"The American Economic Keuietu (May, 1984). pp. 116-121.TVersky, A., and D. Kahneman, "Rational Choice and the framing of Decisions," The Journalof Business (October, 1986). pp. 251-278.Theil, H., Principles of Econometrics (John Wiley & Sons Inc., 1971).Utzlg, L. J. , "CMS Performance Measurement. Ch apter 6" In Cost Mana gement Jor Today'sAdvanced Manufacturing. C. Berliner and J. A. Brimson, E ds. (Harvard Bu siness SchoolPress, 1988).Voss. C. A., Just-in-Ttme Manujacture (IFS Publications Ltd, 1987).W ex l^ , K. N.. and G. A. YukI, Organizational Behavior an d Personrwl PsucholoQu (Richard DIrwin Inc., 1984).Womaek, J. P., D. T. Jones, and D. Ross, The Machir\e that Changed the World (RawsonAssociates, 1990).Young. S. M.. M. D. Shields, and G. Wolf. "Manufacturing Controls and Performance: AnExperiment." Accounting. OrgarUzations. an d Society (13. 1988), pp. 607 -618 .Zipkin, P. H., "Does Manufacturing Need a JIT Revolution?" Harvard Business Review 69(Janu ary-Februa iy 1991), pp. 40-49 .


24/24

Documents

27Reporting Manufacturing Porformance Measures