Strategy, Choice of Performance Measures, and Performance

Preparing to load PDF file. please wait...

0 of 0
100%
Strategy, Choice of Performance Measures, and Performance

Transcript Of Strategy, Choice of Performance Measures, and Performance

BEHAVIORAL RESEARCH IN ACCOUNTING Volume 18, 2006 pp. 185–205
Strategy, Choice of Performance Measures, and Performance
Wim A. Van der Stede University of Southern California
Chee W. Chow San Diego State University
Thomas W. Lin University of Southern California
ABSTRACT: We examine the relationship between quality-based manufacturing strategy and the use of different types of performance measures, as well as their separate and joint effects on performance. A key part of our investigation is the distinction between financial and both objective and subjective nonfinancial measures. Our results support the view that performance measurement diversity benefits performance as we find that, regardless of strategy, firms with more extensive performance measurement systems—especially those that include objective and subjective nonfinancial measures—have higher performance. But our findings also partly support the view that the strategy-measurement ‘‘fit’’ affects performance. We find that firms that emphasize quality in manufacturing use more of both objective and subjective nonfinancial measures. However, there is only a positive effect on performance from pairing a qualitybased manufacturing strategy with extensive use of subjective measures, but not with objective nonfinancial measures.
INTRODUCTION
Performance measures play a key role in translating an organization’s strategy into desired behaviors and results (Campbell et al. 2004; Chenhall and Langfield-Smith 1998; Kaplan and Norton 2001; Lillis 2002). They also help to communicate expectations, monitor progress, provide feedback, and motivate employees through performancebased rewards (Banker et al. 2000; Chenhall 2003; Ittner and Larcker 1998b; Ittner et al. 1997; Ittner, Larcker, and Randall 2003). Traditionally, firms have primarily used financial measures for these purposes (Balkcom et al. 1997; Kaplan and Norton 1992). But with the ‘‘new’’ competitive realities of increased customization, flexibility, and responsiveness, and associated advances in manufacturing practices, both academics and practitioners have argued that traditional financial performance measures are no longer adequate for these functions (Dixon et al. 1990; Fisher 1992; Ittner and Larcker 1998a; Neely 1999). Indeed, many
We acknowledge the helpful suggestions by Tom Groot, Jim Hesford, Ranjani Krishnan, Fred Lindahl, Helene Loning, Michal Matejka, Ken Merchant, Frank Moers, Mark Peecher, Mike Shields, Sally Widener, workshop participants at the University of Illinois, the 2002 AAA Management Accounting Meeting in Austin, the 2002 World Congress of Accounting Educators in Hong Kong, and the 2003 AAA Annual Meeting in Honolulu. An earlier version of this paper won the best paper award at the 9th World Congress of Accounting Educators in Hong Kong (2002).
185

186

Van der Stede, Chow, and Lin

accounting researchers have identified the continued reliance on traditional management accounting systems as a major reason why many new manufacturing initiatives perform poorly (Banker et al. 1993; Ittner and Larcker 1995).
In light of this development in theory and practice, the current study seeks to advance understanding of the role that performance measurement plays in executing strategy and enhancing organizational performance. It proposes and empirically tests three hypotheses about the performance effects of performance measurement diversity; the relation between quality-based manufacturing strategy and firms’ use of different types of performance measures; and the joint effects of strategy and performance measurement on organizational performance.
The distinction between objective and subjective performance measures is a pivotal part of our investigation. Prior empirical research has typically only differentiated between financial and nonfinancial performance measures. We go beyond this dichotomy to further distinguish between nonfinancial measures that are quantitative and objectively derived (e.g., defect rates), and those that are qualitative and subjectively determined (e.g., an assessment of the degree of cooperation or knowledge sharing across departmental borders). Making this finer distinction between types of nonfinancial performance measures contributes to recent work in accounting that has begun to focus on the use of subjectivity in performance measurement, evaluation, and incentives (e.g., Bushman et al. 1996; Gibbs et al. 2004; Ittner, Larcker, and Meyer 2003; MacLeod and Parent 1999; Moers 2005; Murphy and Oyer 2004).
Using survey data from 128 manufacturing firms, we find that firms with more extensive performance measurement systems, especially ones that include objective and subjective nonfinancial measures, have higher performance. This result holds regardless of the firm’s manufacturing strategy. As such, our finding supports the view that performance measurement diversity, per se, is beneficial.
But we also find evidence that firms adjust their use of performance measures to strategy. Firms that emphasize quality in manufacturing tend to use more of both objective and subjective nonfinancial measures, but without reducing the number of financial measures. Interestingly, however, combining quality-based strategies with extensive use of objective nonfinancial measures is not associated with higher performance. This set of results is consistent with Ittner and Larcker (1995) who found that quality programs are associated with greater use of nontraditional (i.e., nonfinancial) measures and reward systems, but combining nontraditional measures with extensive quality programs does not improve performance. However, by differentiating between objective and subjective nonfinancial measures—thereby going beyond Ittner and Larcker (1995) and much of the extant accounting literature—we find that performance is higher when the performance measures used in conjunction with a quality-based manufacturing strategy are of the subjective type.
Finally, we find that among firms with similar quality-based strategies, those with less extensive performance measurement systems have lower performance, whereas those with more extensive performance measurement systems do not. In the case of subjective performance measures, firms that use them more extensively than firms with similar qualitybased strategies actually have significantly higher performance. Thus, a ‘‘mismatch’’ between performance measurement and strategy is associated with lower performance only when firms use fewer measures than firms with similar quality-based strategies, but not when they use more.
The paper proceeds as follows. The next section builds on the extant literature to formulate three hypotheses. The third section discusses the method, sample, and measures.

Behavioral Research in Accounting, 2006

Strategy, Choice of Performance Measures, and Performance

187

The fourth section presents the results. The fifth section provides a summary, discusses the study’s limitations, and suggests possible directions for future research.
HYPOTHESES Although there is widespread agreement on the need to expand performance measurement, two different views exist on the nature of the desirable change (Ittner, Larcker, and Randall 2003; Ruddle and Feeny 2000). In this section, we engage the relevant literatures to develop three hypotheses. Collectively, the hypotheses provide the basis for comparing the two prevailing schools of thought on how performance measurement should be improved; that of performance measurement diversity regardless of strategy versus that of performance measurement alignment with strategy (Ittner, Larcker, and Randall 2003).
The Performance Measurement Diversity View
A number of authors have argued that broadening the set of performance measures, per se, enhances organizational performance (e.g., Edvinsson and Malone 1997; Lingle and Schiemann 1996). The premise is that managers have an incentive to concentrate on those activities for which their performance is measured, often at the expense of other relevant but non-measured activities (Hopwood 1974), and greater measurement diversity can reduce such dysfunctional effects (Lillis 2002). Support for this view is available from economicsbased agency studies. Datar et al. (2001), Feltham and Xie (1994), Hemmer (1996), Holmstrom (1979), and Lambert (2001), for example, have demonstrated that in the absence of measurement costs, introducing incentives based on nonfinancial measures can improve contracting by incorporating information on managerial actions that are not fully captured by financial measures. Analytical studies have further identified potential benefits from using performance measures that are subjectively derived. For example, Baiman and Rajan (1995) and Baker et al. (1994) have shown that subjective measures can help to mitigate distortions in managerial effort by ‘‘backing out’’ dysfunctional behavior induced by incomplete objective performance measures, as well as reduce noise in the overall performance evaluation.
However, the literature also has noted potential drawbacks from measurement diversity. It increases system complexity, thus taxing managers’ cognitive abilities (Ghosh and Lusch 2000; Lipe and Salterio 2000, 2002). It also increases the burden of determining relative weights for different measures (Ittner and Larcker 1998a; Moers 2005). Finally, multiple measures are also potentially conflicting (e.g., manufacturing efficiency and customer responsiveness), leading to incongruence of goals, at least in the short run (Baker 1992; Holmstrom and Milgrom 1991), and organizational friction (Lillis 2002).
Despite these potential drawbacks, there is considerable empirical support for increased measurement diversity. For example, in a study of time-series data in 18 hotels, Banker et al. (2000) found that when nonfinancial measures are included in the compensation contract, managers more closely aligned their efforts to those measures, resulting in increased performance. Hoque and James (2000) and Scott and Tiessen (1999) also have found positive relations between firm performance and increased use of different types of performance measures (e.g., financial and nonfinancial). These results are consistent with nonfinancial performance measures containing additional information not reflected in financial measures.
Although prior studies like these have advanced our understanding of performance measurement design, they share the limitation of not distinguishing among measures that are determined objectively and ones that are based on subjective judgment. A key difference between objective and subjective measures is that the latter are often less accurate and

Behavioral Research in Accounting, 2006

188

Van der Stede, Chow, and Lin

reliable, and are more open to raters’ biases (Campbell 1990; Fulk et al. 1985; Hawkins and Hastie 1990; Heneman 1986). Prendergast and Topel (1993), for example, have suggested that allowing subjective judgment in performance evaluation can reduce employee motivation. This result is due to the latitude for evaluators to ignore performance measures that are included in the performance plan, and to use measures that differ from those originally planned. Moreover, when evaluations are subjective, employees may divert job effort toward influencing their supervisors’ evaluations (Milgrom 1988; Prendergast 1993; Prendergast and Topel 1996). Whether limitations like these outweigh the benefits of subjective measures is an important question.
To date, only a few studies have empirically examined the use and effects of subjectivity in performance measurement. Of particular relevance to the current study, Ittner, Larcker, and Meyer (2003) analyzed how subjective performance measures were used in a leading international financial services provider. They found that leaving room for subjectivity allowed supervisors to ignore many performance measures, and short-term financial measures became the de facto determinants of bonus awards. Employee dissatisfaction with the system was so high that the firm reverted to basing bonuses strictly on revenue. Moers (2005) also found that the use of subjective performance measures induces performance evaluation bias. In a study of managerial compensation plans in a privately owned Dutch maritime firm, he found that the use of multiple objective and subjective performance measures was related to more compressed and more lenient performance ratings. These findings by Ittner, Larcker, and Meyer (2003) and Moers (2005) are useful for understanding the uses and limitations of subjective performance measures; however, their generalizability is limited by each study having focused on only one company.
In sum, prior studies have argued for performance measurement to extend beyond just financial measures, and importantly for this study, a case has been made for differentiating between objective and subjective nonfinancial measures. Although arguments have been proffered about the downsides of performance measurement diversity, in general, and about the inclusion of subjective performance measures, in particular, the expectation is that:
H1: Firm performance will be positively associated with performance measurement diversity, particularly when the performance measurement system is extended to include (more) objective and subjective nonfinancial performance measures.
The Performance Measurement Alignment View
In contrast to proponents of performance measurement diversity, contingency theory maintains that the optimal design of performance measurement systems depends on the organization’s strategy (and other organizational characteristics), and that performance will be higher only if both are aligned (for reviews, see Chenhall 2003; Fisher 1995; LangfieldSmith 1997). In this study, we specifically consider the performance measurement implications of quality-based manufacturing strategy.
The primary reason for this focus is that quality-based initiatives have challenged the relevance of traditional financial measures (Abernethy and Lillis 1995; Dixon et al. 1990), and ‘‘one way in which this challenge has been met is by performance measurement system expansion’’ (Lillis 2002, 498). For example, nonfinancial measures can help ensure that quality improvement results, which often take considerable time to materialize, are not subjugated to financial results (Daniel and Reitsperger 1991; Ittner and Larcker 1995). Moreover, some of the key dimensions of quality-focused strategies, such as those focused

Behavioral Research in Accounting, 2006

Strategy, Choice of Performance Measures, and Performance

189

on knowledge sharing and cooperativeness (Lillis 2002), are difficult to quantify and, thus, may need to be assessed subjectively.
Prior studies in this area, however, have not distinguished between objective and subjective nonfinancial performance measures. Moreover, although some prior studies have found associations between the choice of performance measures and the type of strategy pursued (e.g., Abernethy and Lillis 1995; Daniel and Reitsperger 1991; Ittner and Larcker 1995; Perera et al. 1997), they have rarely investigated the performance effects of such choices. Or, as Ittner and Larcker (1998a, 221) and Chenhall (2003, 142) have emphasized, the scant empirical evidence that has been reported on the strategy-measure-performance relationship—only some of which relates to manufacturing strategy—is equivocal at best. In sum, considering the limited evidence in conjunction with the failure of prior work to distinguish between objective and subjective performance measures, there is room for further investigation. Next, we review related prior studies as the basis for developing two hypotheses.
Abernethy and Lillis (1995) found that reliance on traditional (cost efficiency-based) measures was positively associated with performance for firms pursuing a nonflexible manufacturing strategy, and negatively associated with performance for firms pursuing a flexible manufacturing strategy. However, their study did not examine the performance of firms that combine nontraditional performance measures with a flexible manufacturing strategy. Furthermore, discriminatory power was low because most firms in the study extensively used virtually all of the performance measures.
Ittner and Larcker (1995) focused on firms using advanced quality programs (TQM). Their sample consisted of 249 firms from the automobile and computer industries in four countries. Using data from a consulting firm, they found ‘‘no support for the proposition that, holding other determinants of performance constant, the highest performance levels should be achieved by organizations making the greatest use of both TQM practices and nontraditional information and reward systems’’ (Ittner and Larcker 1995, 2). On the contrary, among firms with extensive use of advanced quality programs, those with a strong reliance on nontraditional performance measures had lower performance than those with less extensive use of such measures. In another study based on the same sample, Ittner and Larcker (1997) examined the relation between quality-focused strategies and strategic control practices, including the importance placed on quality performance in determining managerial compensation. The performance effects (on ROA, ROS, sales growth, and selfreported performance) of a match between the two were generally insignificant.
While the findings of both their 1995 and 1997 studies are informative, Ittner and Larcker (1995) caution that both industries in their sample may be so competitive that there was insufficient variation in the variables for identifying significant performance effects. A further limitation is that the studies only had access to measures of the overall importance placed on nonfinancial and quality performance. There was no information about the kinds and numbers of measures used of each type, nor was it possible to distinguish between objective and subjective nonfinancial performance measures.
Said et al. (2003) also examined the fit between operational and competitive circumstances and firms’ choice of performance measures, as well as the performance effects of including nonfinancial measures in compensation contracts. A notable advance over prior studies is the use of a large sample (1,441 firm-years for firms using nonfinancial performance measures in managerial bonus plans, matched against the same number of firm-years for firms using exclusively financial measures). The results indicated that firms with a greater quality focus made greater use of nonfinancial measures. Furthermore, consistent

Behavioral Research in Accounting, 2006

190

Van der Stede, Chow, and Lin

with the contingency-based approach to performance measurement, firms that relied on nonfinancial measures either more or less than a benchmark model had lower performance. Notably, these performance effects diverged from those of Ittner, Larcker, and Randall (2003) where negative deviations from the benchmark had no detectible effect and positive deviations were associated with higher performance. However, Said et al. (2003) only used crude indicators (either a dummy variable or the total weight) to reflect a firm’s use or nonuse of nonfinancial measures and, hence, like the other studies of quality initiatives, they did not distinguish among different types of nonfinancial performance measures.
Taken as a whole, the contingency theory-based studies have made a case for the need to align performance measures with quality strategy. Yet they have produced only limited, and at times mixed, findings on the relation between quality initiatives and performance measure use and their joint effect on performance. Our next two hypotheses provide focus to further empirical investigation and extend prior work by incorporating the distinction between objective and subjective nonfinancial measures.
H2: Firms that place more emphasis on quality in the manufacturing strategy will extend their performance measurement system to include (more) objective and subjective nonfinancial performance measures.
H3: Firms that place more emphasis on quality in the manufacturing strategy will have higher performance only when they match their performance measurement system to their strategy, that is, when they extend their performance measurement system to include (more) objective and subjective nonfinancial performance measures.
Together, these two hypotheses posit that firms pursuing a quality-based manufacturing strategy will tend to use more extensive performance measurement systems that include objective and subjective nonfinancial performance measures, and when they do—thus aligning their choice of performance measures with their manufacturing strategy—they will exhibit superior performance.
METHOD Sample
We employ the survey method because the data sought are not available from public sources.1 Given the central role of manufacturing strategy in our study, we target managers or directors of manufacturing in manufacturing firms.
To obtain a large enough sample for statistical tests, we include both U.S. and European firms. Using the Explore database from CorpTech, we restrict the U.S. sample to Southern California, where the authors’ schools are likely to have some goodwill. We select manufacturing firms with at least 50 employees and U.S.$2.5 million in annual sales (678 firms) to ensure that the sample firms would have sufficient sophistication and scale to use advanced management practices (performance measurement systems, in particular). We

1 Based on recommendations by the Total Design Method (Dillman 1978, 1999) and other sources on survey design (see Van der Stede et al. [2005] for a review), the authors went through several iterations of the survey before submitting it to the scrutiny of four management accounting faculty colleagues for pre-testing. We also searched the literature for existing scales whenever available. Finally, we did follow-ups and conducted nonresponse bias analyses (see below).

Behavioral Research in Accounting, 2006

Strategy, Choice of Performance Measures, and Performance

191

received 106 replies, 19 of which were unusable for a range of reasons, leaving a usable sample of 87.2
In Europe, we geographically restrict our target sample to Belgium where we partnered with the Vlerick Leuven Ghent Management School (VLGMS), the premier Belgian business school. The VLGMS maintains a database of Belgian firms from which we select a sample using the same employment and sales criteria as in the U.S. We obtained 48 replies from a sample of 447 firms. Of these, 41 were usable.3
Hence, our total usable sample size is 128 (87 ϩ 41). The sample firms have an average headcount in the manufacturing department (firm as a whole) of about 227 (3,216). Their average annual production value (sales) in the manufacturing department (firm as a whole) is about U.S.$86 million (U.S.$1.2 billion).4 On average, respondents have been working for their current firms for about 11 years, and in their current positions for 6 years. Their average age is 46 years, and most have a college degree. Of the above firm and respondent characteristics, only the respondents’ years at the firm (10 versus 13) and age (47 versus 41) are significantly different between the U.S. and Belgian samples, respectively (p Ͻ 0.10). We can discern no reason why these two characteristics would bias the results. More importantly, the U.S. and Belgian samples do not differ (p Ͼ 0.10) along any selfreported performance dimension (financial, operating, employee-related, and customerrelated).
We perform two tests to assess the potential for nonresponse bias. First, we compare sales and employment (available in the databases) between the responding and nonresponding firms. Neither variable is significantly different between U.S. and Belgian respondents and nonrespondents. Second, we compare early versus late respondents (i.e., those who responded after the final follow-up) on all the variables used in the analyses (quality-based manufacturing strategy, number of each category of measures, and performance). We find no significant difference (at p Ͻ 0.10) on any of these variables in either the U.S. or Belgian sample. While these tests do not constitute definitive evidence, they do increase our confidence that there is neither significant nonresponse bias between the target and survey samples, nor important systematic differences between the U.S. and Belgian samples.
Variables
The survey contains questions on the performance measures in use, manufacturing strategy, performance, and the firm and respondent demographics reported above.
2 The 106 replies represent 16 percent of the 647 valid mailings (31 of the 678 surveys were returned as undeliverable). Of the 106 replies, 19 were either partially incomplete, or uncompleted for reasons as diverse as: addressee deceased or no longer with the firm (7); interim manager (1); no time (2); firm in bankruptcy or reorganization (2); firm in start-up (1); nonmanufacturing (1); nondomestic operations (1); and policy of nonparticipation (4). Of the 87 usable responses (106 – 19), we received 38 (44 percent) immediately without follow-up, 20 (23 percent) after the first follow-up without replacement, and 29 (33 percent) after the second follow-up with replacement.
3 The 48 replies represent 11 percent of the 438 valid mailings (9 of the 447 surveys were returned as undeliverable). Of the replies, 7 were not usable because operations had been stopped (3); the company had a policy of nonparticipation (3); and one addressee promised to respond later, but never did. Of the 41 usable responses (48 – 7), we received 27 (66 percent) immediately without follow-up and 14 (34 percent) after the first and only follow-up with replacement. We did not do an intermediate follow-up (just a reminder letter) in Belgium as we did in the U.S. because of cost considerations and (international) administrative complexity.
4 The rather large difference between the company-wide and manufacturing-specific statistics is due to 45 percent of the manufacturing departments being part of a profit center, division, or business unit within the company. For the other 55 percent of the manufacturing departments in our sample, the next higher level is the firm as a whole.

Behavioral Research in Accounting, 2006

192

Van der Stede, Chow, and Lin

Performance Measures in Use Because of its relative complexity, we reproduce the performance measurement sec-
tion of the survey in the Appendix. This section contains one subsection on financial measures, three subsections on objective nonfinancial measures (internal operating measures, employee-oriented measures, and customer-oriented measures), and one subsection on subjective performance measures.5 We asked participants to indicate the specific performance measures currently used by upper management in their firms for performance measurement and evaluation of the manufacturing department.6 Respondents could check off measures from the list and also write in additional measures. Table 1 reports the distributional statistics of these data.
The data reveal a wide range across the sample firms in the total number of performance measures used. We take this to imply that the respondents had tried to reflect the actual situations in their firms, and not simply checked off items on the list. Another indication that the respondents were seriously engaged is that 28 out of 128 respondents (22 percent) took the effort to write in a total of 38 additional measures.7 Moreover, more than half of the respondents indicated that their firms had made changes to their performance measurement systems in the past three years, which we view as evidence that the firms consider issues of performance measurement important enough to merit continual scrutiny and adjustment.8
Table 1 shows that among the sample firms, the average performance measurement system contains about 25 percent financial measures, 56 percent objective nonfinancial measures, and 19 percent subjective measures. More specifically, the most numerous are internal operating measures (26 percent), followed by financial measures (25 percent), and

5 We obtained this list of measures from a wide reading of both the academic and practitioner literatures, including Abernethy and Lillis (1995), Butler et al. (1997), Chow et al. (1997), Clinton and Hsu (1997), Epstein and Manzoni (1997), Fisher (1992), Hoffecker and Goldenberg (1994), Maisel (1992), Kaplan and Norton (1992, 1993, 1996a, 1996b, 1996c, 2000, 2001), Provost and Leddick (1993), Scott and Tiessen (1999), and Stivers et al. (1998).
6 Similar to Ittner and Larcker (1995), we consider ‘‘use’’ to mean ‘‘the existence of a measure in the performance measurement and evaluation system’’ without getting into whether the specific measure is used for informational or reward purposes, or both. To curb a potential upward bias in the number of measures reported—i.e., situations where financial, objective nonfinancial, or subjective performances are tracked but not used (Stivers et al. 1998)— we explicitly direct respondents ‘‘to only check (or write in) those measures that are reported, analyzed, and discussed on a regular basis for the purpose of performance measurement and evaluation’’ (see the Appendix).
7 Of these 28 respondents, 21 provided one write-in measure, six provided two write-in measures, and one wrote in a total of five measures. By measurement category, there are 13 financial, 19 objective nonfinancial, and six subjective write-ins. Some are idiosyncratic, such as my ability to work cooperatively with the Japanese top management. Less idiosyncratic measures are capital expenditures and receivables on net sales for financial; order backlog, product return turnaround cycle time, and quoted lead time accuracy for objective nonfinancial; and my ability to maintain strong internal control for subjective measures.
8 Twenty-five percent of the respondents reported that their firms’ performance measurement system had been modified in the last year, 34 percent in the last 2–3 years, and 41 percent reported no change in the last three years. The percentage of sample firms indicating a change in the performance measurement system in the last three years (59 percent) is almost identical to that reported in a survey by the Institute of Management Accountants (1996) (60 percent). Of the firms in our sample reporting a change during the last 3 years, the change involved fine-tuning financial measures in use (53 percent); dropping some financial measures (11 percent); adding financial measures (47 percent); fine-tuning nonfinancial measures in use (43 percent); dropping some nonfinancial measures (5 percent); or adding nonfinancial measures (64 percent). These percentages sum to over 100 percent, indicating that the changes are multidimensional. They also reveal that adding both nonfinancial and financial measures are common, with the former being more prevalent (64 percent versus 47 percent). Moreover, dropping nonfinancial measures (5 percent) is less common than dropping financial measures (11 percent). Overall, the direction of change seems to be toward greater use of nonfinancial performance measures.
Behavioral Research in Accounting, 2006

Strategy, Choice of Performance Measures, and Performance

193

TABLE 1 Performance Measurement System Composition
(n ‫ ؍‬128)

Min Max

Number of financial measures (# FIN)a

1

11

Number of objective nonfinancial measures (# NONFIN)a 0

26

Number of internal operating measures (# OPS)a

0

11

Number of employee-oriented measures (# EMPL)a

0

8

Number of customer-oriented measures (# CUST)a

0

8

Number of subjective measures (# SUBJ)a

0

7

Total number of measures (# TOTAL)a

6

41

Mean
6 14 6
4 4 5
25

Std.Dev.
3 5 2 2 2 2
7

Percent financial measures (% FIN)b

0.08 1.00

0.25

0.09

Percent objective nonfinancial measures (% NONFIN)b

0.00 0.83

0.56

0.09

Percent internal operating measures (% OPS)b

0.00 0.67

0.26

0.07

Percent employee-oriented measures (% EMPL)b

0.00 0.33

0.15

0.06

Percent customer-oriented measures (% CUST)b

0.00 0.27

0.15

0.06

Percent subjective measures (% SUBJ)b

0.00 0.43

0.19

0.08

100%

Categorical measure of measurement diversity (C DIVERSITY)c

1.00 4.00

3.29

0.99

a See the Appendix for the list of performance measures and their breakdown by category (financial, internal
operating, employee-oriented, customer-oriented, and subjective). b As a percentage of the total number of performance measures used in each firm. c The categorical measure of performance measurement C DIVERSITY has the following values when the unit
uses: 1 ϭ financial measures only (i.e., measures checked in Section 1 of the Appendix only); 2 ϭ financial and objective nonfinancial measures (i.e., measures checked in Sections 1 and 2 only); 3 ϭ financial and subjective measures (i.e., measures checked in Sections 1 and 3 only); and 4 ϭ financial and objective nonfinancial and subjective measures (i.e., measures checked in Sections 1, 2, and 3).

lower but still non-negligible proportions of subjective performance measures (19 percent), employee-oriented measures (15 percent), and customer-oriented measures (15 percent).9
We operationalize the notion of performance measurement ‘‘diversity’’ using two different approaches (see Table 1). Our first operationalization is a count measure, that is, the number of financial (# FIN), objective nonfinancial (# NONFIN), and subjective (# SUBJ) measures. Our second operationalization is a categorical measure (C DIVERSITY) where 1 ϭ financial measures only (i.e., measures checked in Section 1 of the Appendix only); 2 ϭ financial and objective nonfinancial measures (i.e., measures checked in Sections 1 and 2 only); 3 ϭ financial and subjective measures (i.e., measures checked in Sections 1 and 3 only); and 4 ϭ financial and objective nonfinancial and subjective measures (i.e., measures checked in Sections 1, 2, and 3). This categorical approach facilitates recognizing another dimension of performance measurement diversity, as in the case of comparing a firm that has two each of the three types of measures versus one that has three measures each from only two categories. Moreover, objective nonfinancial measures
9 The distribution of measures across measurement types (financial, objective nonfinancial, and subjective) is not significantly different between the U.S. and Belgian samples. Hence, for parsimony, we only report the results based on the combined sample. Moreover, when we include a dummy variable for the sample firms’ national origin in the subsequent tests, we find it to be uniformly insignificant, with no qualitative impact on the results.
Behavioral Research in Accounting, 2006

194

Van der Stede, Chow, and Lin

such as productivity and other efficiency measures are in some sense ‘‘proxies’’ for financial measures (e.g., they are more timely measures, and thus, leading proxies for financial measures), whereas subjective measures add an extra dimension to performance measurement. Our C DIVERSITY measure reflects this distinction by differentiating between measurement systems that use financial and objective nonfinancial measures (ϭ2) versus those that use financial and subjective measures (ϭ3).
Manufacturing Strategy We derive our measure of quality-based manufacturing strategy primarily from
Abernethy and Lillis (1995), Chenhall (1997), Ittner and Larcker (1995), and Perera et al. (1997), supplemented by a wider reading of the academic and practitioner literatures (Banker et al. 1993; Hackman and Wageman 1995; Saraph et al. 1989). In total, the quality strategy scale includes the seven items shown in Table 2. These items capture employee involvement, process improvements, and cross-departmental coordination, which have been

TABLE 2 Measures of Quality Strategy and Performance
(n ‫ ؍‬128)

Quality Strategy (QUALSTRAT) (Eigenvalue ϭ 3.01; Cronbach ␣ ϭ 0.77; ␮ ϭ 3.55, ␴ ϭ 0.63)
In your department, to what extent:a
1. Are nonmanagement employees evaluated for quality performance? 2. Do nonmanagement employees participate in quality improvement
decisions? 3. Is building awareness about quality among non-management
employees ongoing? 4. Are quality performance data displayed at employee work stations /
areas? 5. Are suggestion programs for quality improvement among non-
management employees used? 6. Are programs in place to improve cycle-times (e.g., by reducing
time-delays or non-value-added activities in manufacturing)? 7. Are programs in place to coordinate quality improvements with
other departments within the organization?

Factor Loading
0.67 0.70
0.73
0.56
0.70
0.64
0.57

Alpha if Item Deleted 0.75 0.74
0.73
0.76
0.73
0.74
0.75

Performance (ALLPERF) (Eigenvalue ϭ 2.30; Cronbach ␣ ϭ 0.74; ␮ ϭ 3.67, ␴ ϭ 0.64)
Please rate the following performances of your department (relative to the industry average):b
1. Financial performance of your department? 2. Operating performance of your department? 3. Employee-oriented performance of your department? 4. Customer-oriented performance of your department?

Factor Loading
0.68 0.84 0.73 0.78

Alpha if Item Deleted
0.73 0.62 0.70 0.68

Output of factor analysis with principal component extraction and oblique rotation (␦ ϭ 0). a Anchored as (1) not at all, (2) low extent, (3) medium extent, (4) high extent, and (5) very high extent. b Anchored as (1) well below average; (2) below average; (3) average; (4) above average; and (5) well above
average.

Behavioral Research in Accounting, 2006
MeasuresPerformance MeasuresPerformanceObjectiveIttner