The effect of TQM factors on financial and strategic performance: an empirical test using manufacturing firms.

Barker, Katherine J. ; Emery, Charles R.

ABSTRACT

Although interest in implementing the total quality management (TQM) philosophy as part of competitive and operational strategies has been around for approximately 25 years, there has been little empirical evidence to suggest that it has a positive effect on an organization's financial and strategic performance. Part of the problem is that a testable framework of TQM constructs has been slow to emerge. This research provides a test instrument and an empirically reliable framework to evaluate an organization's TQM implementation. This study of 257 manufacturing firms provides definitive evidence that TQM implementation is a significant predictor of customer satisfaction and a weak but significant predictor that TQM is associated with the improvement of several financial variables. Lastly, the findings suggest that the impact of the TQM variables on performance is highly correlated with duration but is not significantly correlated with either organizational size or industrial specialty. This suggests the robustness of the TQM philosophy as part of any competitive or operational strategy. Lastly, the findings clearly suggest that organizations using the TQM constructs will continue to improve over time.

INTRODUCTION

Total Quality Management (TQM) provides a paradigm shift in management philosophy for improving organization effectiveness (Byrne, 1992; Gagne, 1983; Lowe and Masseo, 1986; Tenner and DeToro, 1992; Waldman, 1994). TQM managers focus efforts of all members to continuously improve all organizational processes and increase value to customers, while relying upon a clear vision of the organization's purpose. This depends on the use of improvement tools (e.g., SPC, benchmarking, process/product improvement teams) and the removal of barriers both within the organization and between the organization and its various stakeholders. TQM has been embraced by thousands of organizations (Lawler and Mohrmon, 1992) as an important management component of operational strategies. However, despite its theoretical promise and enthusiastic response, anecdotal evidence suggests that attempts to implement it and/or achieve financial benefits are often unsuccessful (Erickson, 1992; Fuchsberg, 1992; Kendrick, 1993). Wyatt, a human resources consulting company, surveyed 531 companies that had undergone restructuring in 1992. Only 41% of the 361 companies that started TQM programs as a part of restructuring considered them to have been effective (Fuchsberg, 1993). Similarly, a study by McKinsey & Co. revealed that, of TQM programs in place for more than two years, as many as two-thirds were considered failures by the employees (Doyle, 1992).

Anecdotal studies have most commonly attributed the failures of TQM implementation and financial improvement to deficiencies of: (1) shared vision, (2) application planning, (3) organizational commitment, (4) training, (5) reward systems, (6) empowerment, or (7) cross-functional integration (Brown et al., 1994; Danjin and Cutcher-Gershenfeld, 1992; Doyle, 1992; Emery and Summers, 1992; Gilbert, 1993). Unfortunately, few empirical studies have been initiated to examine relationships between TQM components and the success/failures of implementation. Part of the problem is that an agreed upon framework of TQM constructs has been slow to emerge. Further, few studies have examined performance in relation to the time since the organization began implementation of TQM. Most researchers are in agreement that it takes time to fully adopt most of the TQM initiatives. As such, time must play an important role in the success of a Total Quality Program. For example, the GAO (1991) study found that improving quality is a long-term process and that it took an average of 21/2 years for the companies in their study to realize performance improvements (the range was from 1 to 5 years). Lastly, few researchers have attempted to empirically test TQM organizations for financial benefits and those that have, focused on only one or two variables or constructs. The purpose of this study is first, to develop an aggregated TQM framework from previous studies and second, to use this framework to test the relationship between various constructs/variables (including time since implementation) and financial/strategic performance.

TOTAL QUALITY MANAGEMENT RESEARCH

During the 1990's there were several key studies that attempted to demonstrate a relationship between TQM and organizational performance. Although these studies advanced our qualitative understanding of TQM, they were often flawed by inadequate sample size, failure to fully define constructs, lack of control variables, and for confusing quality management practices with measures of quality performance. For example, The GAO (1991) study suggested that some relationship might exist between TQM practices and improved financial performance, but lacked an adequate number of firms to obtain statistically significant results or any control methodology. The total study included only 20 of the highest scoring Baldrige candidates from 1988 and 1989, and only 15 of these firms were willing to provide financial data. Little distinction was made between quality practices and quality results and only two areas of the Baldrige Award (employee relations and operating procedures) were used to measure the impact of selected portions of TQM on financial performance. In addition to the inadequate sample size, there were no control firms or other control variables present to provide any measure of statistical rigor.

Wisner & Eakins (1994) published a descriptive study of TQM performance using a sample of 17 Baldrige Award winners. Although many non-financial performance measures were reported to have improved for the individual companies, the study was again too small to be statistically valid, and performance measures were not held constant across firms. Financial information was available for only four of the firms that were publicly held. (The others were either privately held or subsidiaries or divisions of other firms.) An attempt was made to show the performance of these four firms against industry averages. However results were mixed. Therefore, due to insufficient data and lack of controls, no statistically valid conclusions could be made.

Madu, Kuei & Lin (1995) selected three quality constructs (customer satisfaction, employee satisfaction and employee service quality) in their attempt to link selected quality practices with improved organizational performance in the United States and Taiwan. Included in their definition of organizational performance were selected financial measures, such as cost performance, profitability and earnings growth. However, it should be noted that all construct and performance indicators were self-reported by survey respondents, and survey respondents can have a tendency to report better than actual results (Bergquist, 1996). The sample sizes were statistically adequate for analysis: 69 U.S. managers, 77 Taiwanese managers, but results differed between the two countries. Managers in Taiwan felt that overall customer satisfaction was the key factor driving improvement in organizational performance, while managers in the U.S. felt that employee satisfaction contributed more to organizational success. No conclusions were presented regarding a relationship between the selected quality practices and improved financial performance.

Easton & Jarrell (1998) were able to show some significant results for firms that had implemented TQM for between 3-5 years. However, it could be argued that the firms included in the sample were those that were just good examples of extremely well-managed firms that used not only TQM, but also JIT, ABC, and re-engineering. There was no objective measurement of specific TQM constructs and inclusion in the sample was solely at the discretion of the interviewer. Financial performance was measured by comparing actual results to analysts' forecasts taken at one point in time for years 1, 2 and 3-5. While analysts' predictions have been used in research when actual information has not been known or has been unavailable, prior research has found that analysts generally have difficulty in forecasting earnings (Jacquillat & Grandin, 1994), that analysts forecasts tend to be overstated (Hunton & McEwen, 1997), and that "some of the most widely followed and recommended stocks have become victims of long-term trends that were widely ignored....including accounting 'irregularities' that were generally unnoticed by analysts" (Regan, 1993). Therefore, the use of actual audited accounting data to establish a change in financial performance provides a more objective basis for observing significant results. Of course many factors not being tested in this study can cause changes in financial performance, including general economic conditions. Attempting to predict what the performance of firms would have been without a total quality program is a difficult, if not impossible, feat. It is very possible that analysts could be biased in their expectations as to anticipated effects of TQM or other micro- or macroeconomic conditions (Hunton & McEwen, 1997) It is not an uncommon occurrence for analysts to miss the mark when quarterly/annual accounting results are released to the financial press. Control portfolios were used to provide a benchmark for non-TQM firms. While attempts were made to find similar firms to those included in the study, it should be noted that the non-TQM control firms were significantly smaller than the sample TQM firms (mean size was less than half that of the TQM firms). Using firms matched by various attributes, since as industry and size, is a common research technique. However each firm is unique and it is extremely difficult to match enough important attributes to be assured that that firms are enough alike to lead to statistically reliable results. Therefore it is questionable as to whether the control firms were appropriately matched to the TQM firms. Using separate variables to control for size and other confounding conditions in a regression would have provided better assurance that an adequate level of control had been established.

Grandzol and Gershon (1997) focused on trying to identify significant relationships between measures of selected TQM practices and indicators of quality achievement, which included financial quality. However, there was no attempt to make an association between TQM (defined as a cohesive set of quality management practices) and improved financial or logistic performance. Therefore the primary research question of interest differed from the current study. One hypothesis specifically tested whether financial quality was a function of three quality practice constructs: continuous improvement, process management and customer focus. However, no statistically significant results could be found between these quality practices and improved financial performance. There was no test of whether all constructs included in their definition of TQM were positively related to improved financial performance. It should also be noted that the financial quality measures (ROI, market share and capital investment ratio) were self-reported and therefore lacked objective evidence. The sample consisted of 275 senior executives from the aerospace, tooling, and engineering industries, which was sufficient to use structural equation modeling (LISREL) to test the hypotheses.

Dixon (1996) used the quality constructs proposed by SB&S (1989) in an attempt to predict levels of financial performance based on the amount of infusion and diffusion of TQM within a firm. Unfortunately, the stated hypotheses were not supported. This may have been because the SB&S (1989) study was one of the first empirical studies to attempt to develop and validate a set of TQM practices (constructs). Subsequent studies have noted that not including a construct related to customer focus or acknowledging the importance of customer satisfaction is a serious shortcoming (Anderson, et al., 1995; Ahire, et al., 1996; Black & Porter, 1996.) An examination of the Baldrige award criteria also shows that practices related to customer focus represent a large percentage of the total points awarded. Therefore, an instrument designed to measure the implementation of TQM practices should include a construct measuring customer focus and involvement in total quality efforts. Additionally, the dependent measures used to test the hypotheses were levels of selected weighted-average financial performance variables rather than changes in these variables. Using a levels variable as a dependent measure does not control for prior performance and is difficult to interpret. The weighting scheme involved using a weight of 4 for the current year, 3 for the previous year, 2 for the second previous year, and 1 for the third previous year. However, no rationale for using this weighting was presented. While Dixon attempted to find a relationship between levels of TQM implementation and financial performance measures, it seems that a more interesting question is whether there is a relationship between TQM implementation and improvement in financial performance. This would involve using changes in financial performance as dependent measures over a specified time period.

Also, during the 1990's there were several key theoretical and empirical studies that proposed similar TQM constructs (Saraph, Benson & Shroeder, 1989; Flynn, Schroeder & Sakakibara, 1994; Anderson, Rungtusanatham & Schroeder, 1994; Flynn, Schroeder & Sakakibara, 1995; Ahire, et al., 1996; Black & Porter, 1996). The following framework of TQM constructs are presented as an amalgamation of these studies along with those practices identified and measured by the Malcolm Baldrige Award criteria (Baldrige, 2005) (see Tables 1 and 2). Additionally, care has been made to include only quality management practices and not performance measures found in some of the studies (Anderson, et al., 1995; Ahire, et al., 1996).

Top Management Commitment

Top management commitment is a necessary and essential element for achieving successful implementation of a total quality program (Deming, 1982; Garvin, 1987; Leonard & Sasser, 1982; Saraph, Benson & Schroeder, 1989; Ahire, et al., 1996). Top management is responsible for setting quality goals and strategies and providing resources to enable implementation of a total quality program. It is the second most heavily weighted item of the Malcolm Baldrige Award criteria (Baldrige, 2005).

* Relative importance given by top management to quality as a strategic issue.

* Extent to which top management views quality as more important than costs or schedules.

* Extent to which top management allocates resources toward efforts to improve quality.

* Extent to which top management accepts responsibility for quality performance.

* Extent to which top management is evaluated for quality performance.

* Extent to which top management has established clear quality goals.

Customer focus

Recent empirical studies agree that TQM cannot exist without a strong customer focus. There must be systems and processes devoted to learning more about customer requirements and improving customer satisfaction (GAO, 1991; Dean & Bowen, 1994; Anderson, et al., 1995; Black & Porter, 1996; Ahire, et al., 1996; Madu, et al., 1995;). A main component of Deming's Chain Reaction (1982) was that improving quality through the firm resulted in better quality products at a lower price that would so satisfy customers that market share would be increased, the company would stay in business and more and more jobs would be provided. Many firms now use customer satisfaction as the final judge of quality (Xerox, Motorola) and offer a 100% satisfaction guarantee. Therefore customer focus should be included when measuring implementation of total quality within a firm. It should also be noted that the 1996 Baldrige Award Criteria gave this construct more weight than any other quality practice. This has since changed, customer and market focus is weighted behind the business results, leadership, and measurement/analysis of information criteria and the same as the other criteria (Baldrige, 2005).

* Extent to which customers are considered the final judge of quality.

* Extent to which customers are encouraged to complete satisfaction surveys.

* Extent to which customer satisfaction survey feedback is made available to managers.

* Extent to which customer input is used to improve product quality.

* Extent to which customer complaints are resolved.

* Extent to which customers are invited to participate in product improvement efforts.

Supplier relationships

Deming (1982) was the first to advocate limiting the number of suppliers and establishing long-term relationships based on quality. By selecting and monitoring suppliers based on non-price selection criteria, research has found an improvement in financial and operational performance (Ittner, Larcker, Nagar & Rajan, 1997). Assuring a reliable source of high-quality parts reduces costs related to inspection of in-coming materials and down time due to defective materials. Effective partnering between manufacturers and suppliers allows the manufacturer to reduce ordering and inventory costs, which are important components of total logistics costs.

* Extent to which suppliers are selected on the basis of quality -vs- price.

* Emphasis on long-term supplier relationships.

* Extent to which suppliers are required to be ISO-9000 certified.

* Extent to which suppliers are limited based on quality.

* Extent to which suppliers are evaluated based on delivery performance.

* Extent to which suppliers are involved in strategic quality planning.

Employee training

Total quality management involves combining concepts and practices drawn from various disciplines (e.g. management, marketing, psychology, engineering, etc.) and is so comprehensive that it requires that all employees receive formal training in total quality concepts and tools to be effective (Ishikawa, 1976; Crosby, 1979; Juran, 1980; Deming, 1982; Ahire, et al., 1996). In addition to training in quality concepts, overall performance and employee satisfaction is enhanced when employees also receive technical and vocational work-skill training which develops additional skills and creates value for both employer and employee (Leonard & Sasser, 1982).

* Extent to which all employees receive quality training.

* Extent to which sufficient resources are available for quality training.

* Extent to which employees receive technical and vocational work-skill training.

* Extent to which employees receive training in statistical process techniques.

* Extent to which employees receive training in problem-solving techniques.

* Extent to which employees receive training to work in teams.

Employee empowerment

One definition of empowerment is "giving workers the training and authority they need to manage their own jobs" (Raiborn, Barfield & Kinney, 1996, p. 49). Ahire et al. (1996) state that "employee empowerment is essential to improve in-process quality control" (p. 31). Empowering employees encourages them to take responsibility for their own work and to be more proactive in finding solutions for problems as they arise. Costs of quality can be reduced by detecting and correcting errors during in-process production rather than after production. Empowered employees are encouraged to prevent and/or detect errors early in the production process rather than relying on final inspections. Therefore, empowerment can lead to significant savings by reducing defects and the need for rework.

* Extent to which employees are responsible for inspecting their own work.

* Extent to which employees are encouraged to find and fix problems.

* Extent to which employees are provided resources to fix problems.

* Extent to which employees are provided technical assistance for solving problems.

* Extent to which employees are rewarded for their ability to solve problems.

Continuous improvement tools

Specific tools are available to provide objective ways of measuring and controlling variation in the production process. These are primarily statistical process control (SPC) methods first advocated by Shewhart (1931) and Deming (1982). Since then many researchers have concurred that SPC is an effective way to improve quality on a continuous basis, particularly for firms just adopting quality initiatives (Garvin, 1986; Flynn, Schroeder & Sakakibara, 1995; Ahire, et al., 1996; Grandzol & Gershon, 1997). Benchmarking should also be included as a continuous improvement tool since it seeks out best practices and products from within the firm or among competitors. The objective data which continuous improvement tools provide should be analyzed and used to keep manufacturing processes under control and determine how the firm can make improvements to its products or processes, thereby always striving for continuous improvement.

* Extent of use of SPC in manufacturing processes.

* Extent to which SPC is considered important in achieving quality goals.

* Extent of use of benchmarking other companies' business practices.

* Extent of use of internal benchmarking.

* Extent to which goals are set based on our benchmarking studies.

* Extent to which error/defect/failure rates are readily available.

* Extent to which progress toward quality-related goals is displayed.

Design and process improvement

The construct of design and process improvement includes tools and practices which manage and control design and production systems to maintain and improve quality throughout the organization. Design and process improvement includes design and control of setup procedures, maintenance and repair (Adam, Herschauer, and Ruch, 1981), zero-defect planning (Crosby, 1979), process improvement through problem analysis (Ishikawa, 1976) and design process control (Grandzol & Gershon, 1997). The Ernst & Young Best Practices Report (1993) found that all process improvement practices proved beneficial to firms at all levels of performance.

* Extent of focus on error or defect prevention rather than inspection.

* Extent of efforts to eliminate non-value-added activities.

* Extent of efforts to simplify the production process.

* Extent of efforts to reduce process cycle time.

* Extent of attention and resources provided for preventative maintenance and repair.

* Extent to which inspection, review or checking of work is automated.

* Extent of final inspection, review or checking.

* Extent to which the design process considers ease of implementation and manufacturability.

* Extent to which product designs are thoroughly reviewed and tested before products are produced.

* Extent to which process design is "fool-proofed" and seeks to minimize the possibility of human errors.

Internal cooperation and open organization

A total quality culture emphasizes cooperative behavior between organizational members (Bushe, 1988; Bossink, Gieskes & Pas, 1993), and encourages sharing information and assisting coworkers to accomplish tasks and solve problems (Waldman, 1994). Leonard & Sasser (1982) observed that the most effective quality programs exhibited open and fluid participation that "cut across traditional organizational boundaries" (p. 168). The following topic areas can be used to measure the extent of internal cooperation and open organization:

* Extent to which everyone works well together.

* Extent to which teamwork is used to solve problems.

* Extent to which interdisciplinary teams are used effectively.

* Extent to which problems are usually solved by managers.

* Extent to which departments seem to be in constant conflict.

* Extent to which hourly employees feel free to communicate with management.

* Extent to which management works well together on all important decisions.

HYPOTHESES

After developing the aggregated TQM framework, the primary purpose of this study is to test the framework and implementation time against several key financial and strategic variables. While previous studies have focused on the more traditional financial measures of ROA and ROE, it can be argued that it is very difficult to examine the preciseness of the relationship between TQM and performance (e.g., increased efficiency) using those aggregate measures. Also, a limitation of the use of a ratio-based dependent variable such as ROA is that TQM often involves improving efficiencies or restructuring, which reduce a firm's asset base. This could cause ROA to grow even though cash flows and a firm's value may actually decline over the same period. To alleviate concerns over bias and provide a more pinpoint analysis of TQM correlations, this study focuses on the change in net income and what caused those changes (e.g., change in sales, change in gross profit margin, or change in operating expenses). Additionally, the secondary purpose of the study was to examine the relationships between the number of years of TQM implementation and several financial and strategic variables. It has been previously suggested that the financial and strategic variables should be positively correlated with the length of time that a TQM philosophy has been in place. As such, the following hypotheses are offered for testing.

H1: The aggregate TQM variable will be positively correlated with a change in net income.

H2: The aggregate TQM variable will be negatively correlated with a change in operating expenses as a percentage of sales.

H3: The aggregate TQM variable will be positively correlated with a change in gross profit margin.

H4: The aggregate TQM variable will be positively correlated with a change in sales.

H5: The aggregate TQM variable will be positively correlated with a change in customer satisfaction.

H6: The number of years of TQM implementation will be positively correlated with change in net income.

H7: The number of years of TQM implementation will be positively correlated with a change in operating expenses.

H8: The number of years of TQM implementation will be positively correlated with a change in gross profit margin.

H9: The number of years of TQM implementation will be positively correlated with a change in sales.

H10: The number of years of TQM implementation will be positively correlated with a change in Customer Satisfaction.

H11: The number of years of TQM implementation will be positively correlated with a change in the aggregate TQM score.

METHOD

Sample Selection and Survey Instrument

This study focuses on U.S. manufacturing firms. In order to employ statistically rigorous research methods, a large-scale mail study was used to obtain a sufficient amount of data. This is an acceptable and feasible method of obtaining data from a large number of firms dispersed over a wide geographic area (Sproull 1995). Many businesses, however, are reluctant to provide financial accounting data (GAO 1991; Wisner and Eakins 1994; Kaynak 1996). As such, the target population chosen from manufacturing firms listed on the Compustat database (SIC 2000--3999 or NAICS 30000-33999). Targeted respondents were high-ranking executives, holding titles such as Vice President of Manufacturing, Vice President of Operations, President, etc. According to Phillips (1981) and Miller and Roth (1994) higher-ranking informants are more reliable sources of information than their lower level counterparts. A listing of 3,640 Compustat firms were cross-referenced against the Dun & Bradstreet Million Dollar Directory (1998) to verify addresses, SIC codes, and to obtain specific names and titles of respondents so that surveys could be sent to a named respondent. Not all corporations were listed in the D & B Million Dollar Directory (1998), which reduced the number of firms to 2,263. The potential pool of respondent firms was further reduced to 1,962 because five years of Compustat data was not available for a number of firms.

Development of Survey Instrument

The survey instrument was developed based on the eight TQM constructs synthesized from previous studies (Table 2). Questions were developed after a thorough review of quality management survey questions found in prior research (Saraph 1989; Flynn et al. 1994; Black and Porter 1996; Ahire et al. 1996). They were also cross-referenced with total quality attributes found in the theoretical literature (Shewhart 1931; Crosby 1979; Juran and Gryna 1980; Deming 1982; Ishikawa 1985; Walton 1986; Feigenbaum 1991). Questions were reviewed, critiqued by other quality researchers and accounting faculty at several universities and went through several rounds of revisions. Careful attention was given to making sure that the wording of each question was clear, concise and described only one construct. Examples of topic areas are included after each construct in the previous section. There were a total of 55 questions relating to quality practices used at each firm. The order of the survey questions was scrambled and some questions were reverse coded to minimize common method variance (Babbie 1990). The survey instrument is available upon request to the senior author.

Confirmatory Factor Analysis

Since this study is based on TQM constructs synthesized from prior empirical research, confirmatory factor analysis was used to validate the constructs. The following tests were performed to assure validity of the constructs and reliability of the survey instrument.

A test of unidimensionality was conducted by examining the survey question correlations. Each question was found to load on the expected construct indicating that all of the constructs are unidimensional. The goodness of fit index (GFI) for each construct ranges from .98 to 1.00, the adjusted goodness of fit index (AGFI) ranges from .94 to .99, and the Bentler-Bonett (1980) normed fit index (NFI) ranges from .94 to .99, all of which indicate that the models fit extremely well and convergent validity is strong. All of the factor loadings representing the observed variables (survey questions) were significant, except for one question relating to supplier relationship and one question relating to internal cooperation and open organizations. These two questions were dropped, which improved the significance of the remaining questions. The theta-deltas, which relate to the error terms of the observed variables, are all significant.

Discriminant validity testing was performed to establish that the constructs are distinctly different from each other. The first step was to examine the correlations between the constructs. These were all quite low, with an average correlation of .21. Next all possible combinations of two constructs are run using two models. In the first model the correlations between the two constructs was allowed to be free. In the second model the correlations between the constructs are fixed at 1.0. If the difference between the Chi-Square statistics was significantly better by allowing the correlations to be free, then there is evidence of strong discriminant validity. In all of the combinations, the Chi-Square statistic significantly deteriorates when the two constructs are forced into one. This shows that each construct is significantly and distinctly different from the other constructs and that discriminant validity has been achieved.

A test of reliability was performed. Cronbach alpha coefficients for this study ranged from 0.89 (continuous improvement tools) to 0.59 (customer focus). Alpha scores that are close to 0.70 or above are considered sufficient for research purposes (Nunnally 1978). The alpha score for the customer focus construct was lower than recommended. This low alpha score may be the result of using only five questions to measure this construct, since adding questions to measure a construct usually raises the alpha score. However, in this study, a strong effort was made to keep the survey as short as possible to increase the response rate. Since this construct is strongly supported in theory and prior research, the customer focus construct is retained. However, including it increases the amount of noise in the TQM variable, and decreases the chances of finding significant results.

A test for non-response bias was accomplished by comparing the mean responses for all the survey questions using the responses received from the 1st mailing as Group 1 (n=117) and the responses received from the 2nd mailing as Group 2 (n=140). Approximately five percent of the questions should be significantly different by chance. Parametric t-tests reveal that only two questions medians are significantly different at the five percent level and four questions are significantly different at the 10 percent level. Non-parametric Wilcoxon scores show two question means were significantly different at the five percent level and five questions were significantly different at the ten percent level. These results are well within the expected levels and indicate that non-response bias is not a problem in this study.

A test of reliability was performed. Reliability refers to the consistency of a measure (Gronlund, 1993). If an instrument is reliable, then it is expected that the scores are an accurate reflection of the respondent's true beliefs. Therefore, if a reliable instrument is administered a second time to the same subjects, their answers should not change from the first administration. There are several theories of reliability and estimates of reliability will differ, to a greater or lesser extent, depending on the specific sources of error being addressed" (Pedhazur & Schmelkin, 1991, p. 88). The most commonly used research method to estimate internal-consistency reliability is the alpha coefficient, alpha often referred to as Cronbach's alpha. (Pedhazur & Schmelkin, 1991). It estimates the reliability of an instrument by measuring the homogeneity of the items in a particular scale. Cronbach alpha coefficients were computed for each construct. The construct alpha scores were similar to those in the Anderson, et al. (1995) study, where alpha coefficients ranged from .60 to .86. The alpha coefficients for this study are as follows:

Continuous improvement tools [alpha] = .89 (11 questions)
Employee training [alpha] = .81 (6 questions)
Top management commitment [alpha] = .79 (6 questions)
Employee empowerment [alpha] = .72 (5 questions)
Design & process improvement [alpha] = .69 (5 questions)
Internal cooperation [alpha] = .68 (11 questions)
Supplier relationship [alpha] = .66 (6 questions)
Customer focus [alpha] = .59 (5 questions)

Alpha scores which are close to .70 or above are considered sufficient for research purposes (Nunnally, 1978). However, it is obvious that one of the constructs, customer focus, has a lower than desirable alpha score (.59). This indicates that the questions addressing customer focus are not reliably measuring the construct. Only five questions were used to measure customer focus, which is the minimum number of questions needed to perform confirmatory factor analysis. If this were an exploratory study that construct would probably be dropped from the definition of TQM. However since these constructs, or components, are derived from both theory and research, the customer focus component of the TQM variable is included in the total definition of TQM. Adding more customer focus questions would very likely have increased the reliability score, but a strong effort was made to keep the survey as short as possible to increase the response rate. Including customer focus is believed to be an essential element of the complete definition of TQM (GAO, 1991; Dean & Bowen, 1994; Anderson, et al., 1995; Madu, et al., 1995; Ahire, et al., 1996; Black & Porter, 1996) and this study is concerned with including all recognized and relevant aspects of TQM. However, including this component with a marginal alpha increases the amount of noise in the TQM variable and therefore decreases the chances of finding significant results. A factor that can have a major impact on the reliability scores are the number of questions which measure a construct. As the number of questions increases the random measurement errors tend to cancel each other out, thus increasing reliability (Brown, 1976). Future studies should try to increase the number of questions measuring each construct to achieve a more reliable measure.

Survey Procedure and Results

The first survey mailing included 1,962 firms. A personalized letter accompanied each survey and the respondents were assured that confidentiality would be maintained. As an incentive to respondents, they were invited to request a Benchmarking Report that provided aggregate information from the study as to their firm, their industry, and all responding manufacturing firms. Reminder postcards were mailed out approximately three weeks after the first mailing. Another letter and copy of the survey were mailed to non-respondents six weeks later. A final reminder postcard was mailed two weeks following the second survey mailing. The number of possible respondents was reduced to 1,810 due to reasons such as surveys returned by the post office as undeliverable, inactive or bankrupt companies, or companies incorrectly identified as manufacturers by Compustat. The number of usable responses received was 257, a response rate of 14.2 percent.

RESULTS

Surprisingly, a large percentage (33%) of the manufacturing firms reported that they had not committed to a formal total quality program. Of the 173 firms that report formal TQM programs, 69% percent (120) report that their program has existed for five or more years. A primary purpose of this study was to develop a framework of TQM constructs and test the framework against two financial and one strategic variable. Hypothesis 1 suggested that the aggregate TQM variable would be positively correlated with change in net income. This hypothesis is supported. The aggregate TQM variable is positively correlated with a change in net income (r=.19 @ p<.01). Regression analysis of the TQM variable indicates that it explains 2.4 percent of the variance in a change in net income at p<.01. Hypothesis 2 suggested that the aggregate TQM variable would be negatively correlated with operating expenses. This hypothesis is supported. The aggregate TQM variable is negatively correlated with operating expenses (r=-.44 @ p<.05). Regression analysis of the TQM variable indicates that it explains 21 percent of the variance in operating expenses at p<.05. Hypothesis 3 suggested that the aggregate TQM variable would be positively correlated with a change in gross profit margin. This hypothesis is not supported. The relationship between the TQM variable and change in gross profit margin is non-significant, i.e. p>.05. Hypothesis 4 suggested that the aggregate TQM variable would be positively correlated with a change in sales. This hypothesis is supported. The aggregate TQM variable is positively correlated with a change in sales (r=.16 @ p<.01). Regression analysis of the TQM variable indicates that it explains 2 percent of the variance in the change of sales at p<.05. Hypothesis 5 suggested that the aggregate TQM variable would be positively correlated to Customer Satisfaction. This hypothesis is supported. The aggregate TQM variable is positively correlated with Customer Satisfaction (r=.39 @ p<.01). Regression analysis of the TQM variable indicates that it explains 15.3 percent of the variance in Customer Satisfaction at p<.01.

A second purpose of this study was to examine the relationship between the number of years of TQM implementation and several financial and strategic variables. Hypothesis 6 suggested that the number of years of TQM (tqmyrs) implementation would be positively correlated with a change in net income. This hypothesis is supported. The TQM years variable is positively correlated with net income (r=.21 @ p<.01). Hypothesis 7 suggested that the number of years of TQM (tqmyrs) implementation would be negatively correlated with operating expenses. This hypothesis is supported. The TQM years variable is negatively correlated with operating expenses (r=.50 @ p<.01). Hypothesis 8 suggested that the number of years of TQM (tqmyrs) implementation would be positively correlated with a change in gross profit margin. This hypothesis is supported. The number of TQM years is positively correlated with a change in gross profit margin (r=.04 @ p<.01). Hypothesis 9 suggested that the number of years of TQM (tqmyrs) implementation would be positively correlated with a change in sales. This hypothesis is supported. The number of years of TQM is positively correlated with a change in sales (r=.18 @ p<.01). Hypothesis 10 suggested that the number of years of TQM (tqmyrs) will be positively correlated to Customer Satisfaction. This hypothesis is supported. The TQM years variable is positively correlated with Customer Satisfaction (r=.42 @ p<.01). Hypothesis 11 suggested that the number of years of TQM implementation would be positively correlated with a change in the TQM aggregate score. This hypothesis is supported. The relationship between the number of years of TQM and the change in the TQM aggregate score is positively correlated (r=.42 @ p<.01).

The third purpose of the study was to examine which of the TQM constructs captured most of the variation of the two financial and one strategic variable. The results of the stepwise regression analysis indicated that "continuous improvement tool" is the best predictor of a change in net income (adjusted [R.sup.2]=.022, F=6.659 @ p<.01). The stepwise regression analysis indicated that internal cooperation and open organization, top management, and customer focus are the best predictors of operating expenses ([R.sup.2]=.068, F=11.417 @ p<.01). The stepwise regression analysis indicated that top management commitment and product improvement are the best predictors of customer satisfaction (adjusted [R.sup.2]=.205, F=34.071 @ p<.01). Each hypothesis was compared by size and by industries within the SIC and the results indicated no significant difference in between industries and size.

DISCUSSION

The primary purpose of this study was to develop and test an unweighted framework of aggregated TQM constructs against several key financial and strategic variables. The hope was that the aggregated TQM variable would be positively correlated with a change in net income. If that was the case, the task was to determine the possible cause of such a change, i.e. a change in sales, a change in gross profit margin, or a change in operating expenses as a percentage of sales. The results indicated that the firms who implemented TQM had a significant increase in net income as a percent of sales. As such, an examination was conducted on the variables that influence net income. As expected there was a positive correlation between a change in sales and the implementation of TQM. However, the regression analysis indicated that TQM was a small predictor ([R.sup.2]=.035 @ p<.05) of the overall variance in sales. Also, the results indicated that there wasn't a significant correlation between the implementation of TQM and a change in gross profit margin (as a percentage of net sales). For the most part, this suggests that while sales were increasing, the cost of goods sold were increasing or decreasing on an inconsistent basis. The most important finding, however, was that operating expenses as a percentage of sales had decreased with the implementation of TQM. And, that the implementation of TQM appeared to be associated with a large part of the variance in operating expenses ([R.sup.2]=.21). This finding is significant and suggests that the implementation of the TQM constructs is associated with a more efficient operation. Also, importantly, the findings indicate a significant relationship between both the TQM aggregate and number of TQM implementation years constructs with customer satisfaction. Additionally, as hoped, both the TQM aggregate and number of TQM implementation years constructs correlate significantly with increased sales.

CONCLUSIONS

This research provides a test instrument and an empirically reliable framework to evaluate an organization's TQM implementation. The research findings provide definitive evidence that TQM implementation is a significant predictor of customer satisfaction and a weak but significant predictor that TQM is associated with the improvement of several financial variables. Further, the regression analysis revealed that "continuous improvement tools" is the best predictor of a change in net income and that "internal cooperation and open organization," "support of top management," and "customer focus" are the best predictors of operating expenses. Also, stepwise regression analysis indicated that "top management support" and "product improvement" are the best predictors of customer satisfaction. Additionally, this research confirms the proposition that the duration of TQM is positively correlated with both financial and strategic variables. Further, the TQM aggregate score (combined factors) gets stronger with age or continued emphasis. Said another way, companies embracing the TQM philosophy should get better and better with continued emphasis on improving the eight component factors. Also, the correlation findings indicate that those companies implementing TQM programs are in fact implementing each of the eight component constructs. This suggests that organizations understand the importance and interrelationship of these constructs. (Note: inter-correlations in Table 3.) Lastly, the findings suggest that the impact of the TQM variables on performance is not significantly different across organizational size and industrial specialty. This is particularly exciting and suggests the robustness of the TQM philosophy.

There are, of course, limitations of this study. It focused entirely on 257 public manufacturing firms within the 2000-3999 SIC and as such, the results cannot be generalized to all firms. Also, while the financial data was collected from the COMPUSTAT data base, the rest of the data was self reported by top management. Future research should use the instrument across a broader base of employees within each organization and attempt to better quantify progress on various strategic variables (e.g., customer satisfaction). Additionally, this study tested the TQM variable with all its component variables weighted equally. Future studies should attempt to use different weighting schemes to test the ability of the TQM variable as a predictor of financial and strategic variables. For example, one could weight the variables similar to those found in the Malcolm Baldrige Award criteria. Also, future studies should consider examining the effectiveness of other strategic initiatives (e.g., reengineering, ISO 9000, Just-in-Time, activity-based costing) in concert with TQM in an effort to explain more of the variance in performance variables. Other considerations might be to create a TQM cultural survey and measure it against performance variables and to examine performance measures five years after the initial performance measurements.

Although interest in implementing the total quality management philosophy as part of an operational strategy has been around for approximately 25 years, it is not any less important today. Regardless of today's competitive strategies, the production cost of goods and services is a key success factor in most industries. Additionally, some measure of customer satisfaction is often a key success factor. As such, a management philosophy such as TQM that reduces operating costs as a percentage of sales and improves customer satisfaction should clearly be part of implementing and maintaining successful functional and operational strategies. The findings of this study clearly indicate that the implementation of TQM can provide improved business performance.

REFERENCES

Adam, E., J. Hershauer, & R.A. Ruch (1981). Productivity and Quality. Englewood Cliffs, NJ: Prentice-Hall.

Ahire, S. L., D.Y. Golhar, & M. A. (1996). Development and validation of TQM implementation constructs. Decision Sciences, 27(1), 23-56.

Anderson, J.C., M. Rungtusanatham, R.G. Schroeder, & S. Devaraj, (1995). A path analytic model of a theory of quality management underlying the Deming management method: Preliminary empirical findings. Decision Sciences, 26(5), 637-658.

Babbie, E. (1990). Survey Research Methods. Belmont, California: Wadsworth Publishing Company.

Baldrige National Quality Program, 2005 (brochure). National Institute of Standards and Technology, Technology Administration, United States Department of Commerce, Gaithersburg, MD.

Bergquist, T.M. (1996). An assessment of the operational and financial impact on companies of quality awards in the United States. Unpublished Dissertation, University of Oregon.

Black, S. A. & L.J. Porter (1996). Identification of the critical factors of TQM. Decision Sciences, 27(1), 1-21.

Bossink, B.A.G., J.F.B Gieskes, & T.N.M. Pas (1992). Diagnosing total quality management--Part I. Total Quality Management, 3(3), 223-231.

Brown, F.A. (1976). Principles of educational and psychological testing (2nd edition). New York, NY: Holt, Rinehart & Winston.

Brown, M.G., D.E. Hitchcock, & M.L. Willard (1994) Why TWM fails and what to do about it. Small Business Reports, 19(7), 58-60.

Bushe, G. R. (1988). Cultural contradictions of statistical process control in American manufacturing organizations. Journal of Management, 14(1), 19-31.

Byrne, J.A. (1992). Management's new gurus. Business Week (August 4), 44-52

Crosby, P. B. (1979). Quality is Free. New York: McGraw-Hill.

Danjin, D. & J. Cutcher-Gershenfeld (1992). Will TQM go the way of QWL? Journal of Quality & Participation, 15(4), 94-97.

Dean, J. W. & D.E. Bowen (1994). Management theory and total quality: Improving research and practice through theory development. Academy of Management Review, 19(3), 392-418.

Deming, W. E. (1982). Out of the Crisis. Cambridge, MA: Massachusetts Institute of Technology.

Dixon, J. M. (1996). Total quality management in ISO-9000 registered organizations: An empirical examination of the critical characteristics associated with levels of financial performance. Unpublished Dissertation: Florida State University.

Doyle, K. (1992) Who's killing total quality? Incentive, 166(8), 12-19.

Easton, G. S. & S.J. Jarrell (1998). The effects of total quality management on corporate performance: An empirical investigation. The Journal of Business, 71(2), 253-307.

Emery, C.R., T.P. Summers, & J.G. Surak (1996). The role of organizational climate in the implementation of TQM. Journal of Managerial Issues, 8(4), 484-496.

Erickson, T.J. (1992) Beyond TQM: Creating the high performance business. Management Review, 5, 58-60.

Ernst and Young and American Quality Foundation (1992). International Quality Study: The Definitive Study of the Best International Quality Management Practices; Top-Line Findings. Ernst and Young, Cleveland, OH.

Feigenbaum, A. V. (1991). Total Quality Control (Third Edition), New York: McGraw-Hill.

Flynn, B. B., R.G. Schroeder, & S. Sakakibara (1994). A framework for quality management research and an associated measurement instrument. Journal of Operations Management, 11(4), 339-366.

Fuchsberg, G. (1992, October 1). Total quality is termed only partial success. The Wall Street Journal, B1

Gagne, J. (1983) America's quality coaches. CPI Purchasing (May), 27-45.

GAO, (1991). Management Practices: U.S. Companies Improve Performance Through Quality Efforts. Washington D.C.: General Accounting Office, GAO/NSIAD-91-190.

Garvin, D. A. (1987). Competing on the eight dimensions of quality, Harvard Business Review, 65(6), 101-109.

Gilbert, G.R. (1993) Employee empowerment: Flaws and practical approaches. Public Manager, 22(3), 45-48.

Grandzol, J. R. & M. Gershon (1997). Which TQM practices really matter: An empirical investigation. Quality Management Journal, 4, 43-50.

Gronlund, N.E. (1993). How to make achievement tests and assessments, Allyn and Bacon, Publishers.

Hunton, J.E. & R.A. McEwen (1997). An assessment of the relation between analysts' earnings forecast accuracy, motivational incentives and cognitive information search strategy. Accounting Review, 72, 497-515.

Ishikawa, K. (1976). Guide to Quality Control. Tokyo: Asian Productivity Organization.

Ittner, C.D., M.V. Rajan, & D.F. Larcker (1997). The choice of performance measures in annual bonus contract. The Accounting Review, 72(2), 231-255.

Jacquillat, B. & P. Grandin (1994). Performance measurement of analysts' forecasts. Journal of Portfolio Management, 21, 94-102.

Juran, J. M. & F.M. Gryna, (1980). Quality Planning and Analysis. New York, McGraw-Hill.

Kaynak, H., (1996). The relationship between just-in-time purchasing and total quality management and their effects on the performance of firms operating in the U.S.: An empirical investigation. Unpublished Dissertation: University of North Texas.

Kendrick, J.J. (1993) TQM: Is it forging ahead or falling behind? Quality, 32(5), 13.

Lawler, E.E. III & S. Mohrmon (1992) Employee Involvement and Total Quality Management: Practices and Results in Fortune 1000 Companies. San Francisco: Jossey-Bass.

Leonard, F. S. & W. Sasser (1982). The incline of quality. Harvard Business Review, 60(5) 163-171.

Lowe, T. & J. Masseo (1986) Three preachers, one religion. Quality (September), 22-25.

Madu, C. N., C.H. Kuei, & C. Lin (1995). A comparative analysis of quality practice in manufacturing firms in the U.S. and Taiwan. Decision Sciences, 26(5), 621-635.

Miller, J. G. & A.V. Roth (1994). A taxonomy of manufacturing strategies. Management Science, 40(3), 285-304.

Nunnally, J. (1978). Psychometric Theory. New York: McGraw-Hill.

Pedhazur, E.J. & L.P. Schmelkin (1991). Measurement, Design, and Analysis: An Integrated Approach, Hillsdale, New Jersey: Lawrence Erlbaum Associates, Inc.

Phillips, L.W. (1981). Assessing measurement error in key information reports: A methodological note on organizational analysis in marketing. Journal of Marketing Research, 18, 395-415.

Raiborn, C., J. Barfield, & M. Kinney (1996). Managerial Accounting (2nd Edition). West Publishing Company.

Regan, P. (1993, July/August). Analyst, analyze thyself. Financial Analysts Journal, 49, 10-12.

Saraph, J. V., P.G. Benson, & R.G. Schroeder (1989). An instrument for measuring the critical factors of quality management, Decision Sciences, 20(4), 810-829.

Shewhart, W.A. (1931). Economic Control of Quality of Manufactured Product. New York: D. van Nostrand.

Sproull, N.L. (1995). Handbook of Research Methods: A Guide for Practitioners and Students in the Social Sciences (2nd Edition), Metuchen, NJ: The Scarecrow Press, Inc.

Tenner, A.R. & I.J. Detoro (1992) Total Quality Managemetn: Three Steps to Continuous Improvement. Reading, MA: Addison-Wesley.

Waldman, D. A.(1994). The contributions of total quality management to a theory of work performance. Academy of Management Review, 19(3), 510-536.

Walton, M. (1986). The Deming Management Method. New York: Putnam Publishing Group.

Katherine J. Barker, SUNY-Fredonia

Charles R Emery, Erskine College

Table 1: TQM Constructs and Sources

A. Saraph et al. (1989)

1. Role of top management
2. Role of quality department
3. Training
4. Product/service design
5. Supplier quality management
6. Process management
7. Quality data and reporting
8. Employee relations

B. Flynn et al. (1994)

1. Top management support
2. Quality information
3. Process management
4. Product design
5. Workforce management
6. Supplier involvement
7. Customer involvement

C. Anderson et al. (1995)

1. Visionary leadership
2. Internal & external cooperation
3. Learning
4. Process management
5. Continuous improvement
6. Employee fulfillment
7. Customer satisfaction

D. Black and Porter (1996)

1. Corporate quality culture
2. Strategic quality management
3. Quality improvement measurement systems
4. People and customer management
5. Operational quality planning
6. External interfacemanagement
7. Supplier partnerships
8. Teamwork structure
9. Customer satisfaction orientation
10. Communication of improvement info

E. Ahire et al. (1996)

1. Top management commitment
2. Customer focus
3. Supplier quality management
4. Design quality management
5. Benchmarking
6. SPC usage
7. Internal quality information usage
8. Employee empowerment
9. Employee involvement
10. Employee training
11. Product quality
12. Supplier performance

F. Baldrige Award Criteria
(Baldrige National Quality
Program, 2005) Points

1. Leadership 120
2. Strategic planning 85
3. Customer and market focus 85
4. Measurement, Analysis, 90
 and Knowledge Management
5. Human Resource Focus 85
6. Process Management 85
7. Business results 450
 1,000

Table 2: TQM Constructs--Current Study

 TQM Constructs--Prior Research and
 Baldrige Award Criteria
TQM Constructs--Current Study (Refer to Table 1 and Key below)

1. Top Management Commitment A(1) B(1) C(1) D(1,2) E(1) F(1,3)
2. Customer Focus B(7) C(7) D(9) E(2) F(6)
3. Supplier Relationships A(5) B(6) C(2) D(7) E(3,12) F(5,7)
4. Employee Training A(3) B(5) C(3) E(10) F(4)
5. Employee Empowerment A(8) B(5) C(6) E(8) F(4)
6. Continuous Improvement Tools A(7) B(2) D(3,10) E(5,6,7) F(2)
7. Design and Process Improvement A(4,6) B(3,4) C(4,5) D(5) E(4) F(5)
8. Internal Cooperation & A(8) B(5) C(2) D(4,6,8) E(9) F(4)
 Open Organization

Table 3: Correlation of TQM factors with Financial and
Strategic Objectives

 1 2 3 4 5

1 Mgt support
2 Customer focus .37 **
3 Supplier .44 ** .49 **
4 Training .50 ** .49 ** .67 **
5 Empowerment .41 ** .59 ** .53 ** .62 **
6 Ci tools .46 ** .54 ** .59 ** .72 ** .53 **
7 Product improve .49 ** .49 ** .47 ** .59 ** .56 **
8 Cooperation .52 ** .47 ** .54 ** .64 ** .54 **
9 Cust Sat .27 ** .21 ** .25 ** .22 ** .20 **
10 Net Income .05 .08 .15 * .13 * .12
11 Ops Expense -.16 ** -.08 -.10 -.11 -.08
12 GPM -.02 -.03 -.04 -.01 -.08
13 Sales .10 .08 .16 ** .09 .13 *
14 Years of TQM .25 ** .22 ** .35 ** .38 ** .26 **
15 TQM aggregate .70 ** .70 ** .75 ** .85 ** .76 **

 6 7 8 9 10

1 Mgt support
2 Customer focus
3 Supplier
4 Training
5 Empowerment
6 Ci tools
7 Product improve .61 **
8 Cooperation .62 ** .57 **
9 Cust Sat .36 ** .24 ** .25 **
10 Net Income .16 ** .14 * .14 * .16 **
11 Ops Expense -.20 ** -.16 * -.23 ** -.16 ** -.42 **
12 GPM -.05 -.05 -.14 .00 .01
13 Sales .15 ** .12 * .14 * .09 .89 **
14 Years of TQM .45 ** .30 ** .36 ** .42 ** .21 **
15 TQM aggregate .83 ** .78 ** .79 ** .39 ** .19 **

 11 12 13 14

1 Mgt support
2 Customer focus
3 Supplier
4 Training
5 Empowerment
6 Ci tools
7 Product improve
8 Cooperation
9 Cust Sat
10 Net Income
11 Ops Expense
12 GPM .05
13 Sales .66 ** .00
14 Years of TQM -.50 ** .04 ** .18 **
15 TQM aggregate -.44 * 0.19 .16 ** .42 **

** correlation is significant at <.01 (2-tailed)

* correlation is significant at <.05 (2-tailed)