The appropriateness of RAPM: toward the further development of theory

Accounting, Organizations and Society 25 (2000) 451±482

www.elsevier.com/locate/aos

The appropriateness of RAPM: toward the further development of theory Frank G.H. Hartmann Faculty of Economics and Econometrics, University of Amsterdam, Roefersstraat 11, 1018 WB, Amsterdam, The Netherlands

Abstract One of the few areas in management accounting and control research that has attained critical mass concerns the organizational and behavioral e€ects of budgeting. Many studies in this area focus on the use of budgets for managerial performance evaluation, and revolve around the construct Reliance on Accounting Performance Measures (RAPM). Despite the volume of the RAPM literature, concern has been expressed about its current state in terms of both theoretical progress and methodological practice. This paper provides an overview of the RAPM literature, and explains some important points of critique. In response to this critique, the last part of the paper explores the possibility for a further development of RAPM theory that is based on the concept of uncertainty. # 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction Within the management accounting and control literature, considerable attention is paid to the behavioral and organizational e€ects of using accounting information for the performance evaluation of subordinate managers. This attention is for a large part directed at describing and understanding the behavioral and organizational e€ects of a construct labeled Reliance on Accounting Performance Measures (RAPM). This construct signi®es: ... the extent to which superiors rely on, and emphasize those performance criteria which are quanti®ed in accounting and ®nancial terms, and which are prespeci®ed as budget targets. (Harrison 1993, p. 319). The RAPM literature has earned a special position in the management accounting literature,

not only because of its volume and thematic constancy (Kren & Liao, 1988; Lau, Low, & Eggleton, 1995), but also because of its impact on other ``streams'' of management accounting research (cf. Chapman, 1997, p. 192). Considering the status of RAPM research, Brownell and Dunk (1991, p. 703) even note (cf. Lau et al., 1995, p. 360): The continuing stream of research devoted to this issue constitutes, in our view, the only organized critical mass of empirical work in management accounting at present. Although these and other claims purport to signal the importance and vitality of the RAPM paradigm, the development of RAPM literature has been accompanied by critical commentaries as well (e.g. Otley, 1980; Briers & Hirst, 1990; Chapman, 1997). In an overview of two decades of RAPM research, Briers and Hirst (1990) point to

0361-3682/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S0361 -3 682(98)00036 -1

452

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

theoretical ¯aws in the RAPM paradigm. Their particular claim is that RAPM research has come to overemphasize statistical sophistication to the detriment of theory development. Critique has also been directed at the research models and research methods typically used in RAPM studies. In this respect, questions have been raised about the uncritical use of contingency theory in RAPM research (Otley, 1980; Schoonhoven, 1981; Chapman, 1997), and about its typical methods for data collection and data analysis. Regarding data collection, for example, Young (1996) points to the often careless use of questionnaire survey methodology, which is the typical data collection method in RAPM studies. Regarding data analysis, Lindsay (1995) has recently criticized the dominance of tests for statistical signi®cance for accepting or rejecting research hypotheses. Therefore, while RAPM studies may form a critical mass in terms of number of papers, the question is left unanswered whether the claim for an organized critical mass is justi®ed. Indeed, the theoretical and empirical ¯aws mentioned impede the development of the theoretically and empirically well grounded body of knowledge, that, almost by de®nition, an organized critical mass of studies should display. The remainder of this paper revolves around this question. It is structured as follows. In the second section, an overview is presented of the origins, the development and the ®ndings of RAPM research. The third section then provides an overview of the methodological and theoretical problems of the RAPM paradigm that challenge the idea of an organized critical mass of RAPM studies. In the fourth section, a way is presented for the further organization of the RAPM paradigm, which involves theory development about RAPM under uncertainty. The ®fth section illustrates the implications of this theory development for the RAPM paradigm, by providing some speci®c directions for further research. The paper closes with some ®nal comments. 2. RAPM research: an overview The starting point of RAPM research lies at the point in time when the human involvement in

budgeting began to receive academic interest (Birnberg, Turopolec, & Young, 1983, p. 116; Briers & Hirst, 1990). A commissioned study by Argyris (1952) suggested the importance of behavioral factors for understanding the e€ectiveness of budgeting. Argyris' ®eld-study of factory supervisors in four production ®rms showed that budgets induced behavioral and organizational e€ects that should be regarded dysfunctional from a management control perspective. The supervisors in this study appeared to perceive budgets as sources of pressure and tension, forcing them to narrow their attention strictly to the problems of their own department. Consequently, the supervisors expressed negative attitudes toward their superiors and toward budget procedures. Argyris' study was important for at least two reasons. First, it showed the need to complement technical knowledge of budgeting with knowledge of human behavior. Second, it suggested that dysfunctional behavior is not just a natural human tendency, causing a need to use controls, but that dysfunctional behavior could indeed be provoked by using (budgetary) controls. Subsequent empirical budgetary research aimed at explaining the incidence of these behaviors rather than at merely describing them. An important step was taken by DeCoster and Fertakis (1968) who investigated the relationship of budget pressure and leadership style for a sample of departmental supervisors. This study contributed to later RAPM research in two respects as well. First, it studied whether budgetpressure would result in an ``initiating structure'' leadership style (undesired), or a ``consideration'' leadership style (desired), and it thus formally addressed the appropriateness of using budgets. Second, their method involved the development of a budget-pressure questionnaire (BPQ). The BPQ was based on role theory (see, e.g., Kahn, Wolfe, Quinn, & Snoek, 1964; Rizzo, House, & Lirtzman, 1970; House & Rizzo, 1972) which, as will be further illustrated below, has become particularly in¯uential in later RAPM studies. It was composed of 97 behaviors regarded indicative of felt budget pressure, about which the authors note: The 97 questions were based upon the hypothesis that the more the supervisor does

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

in response to perceived or actual sent role expectations (i.e., budget requirements) (...), the more pressure he will feel. (DeCoster & Fertakis, 1968, p. 240). Later, Swieringa and Moncur (1972, 1975) used adapted versions of the BPQ to describe and classify, so-called, budget-related behaviors. In this latter and shorter format, the questionnaire has been used in many later RAPM studies for measuring various budget-related variables (e.g. Kenis, 1979; Merchant, 1981, 1984). The early studies, which attempted to broadly map budgetary processes and to explain (dys)functional e€ects of budgeting, were followed by those that focused on discrete steps in the budgeting process. Two steps, corresponding to the beginning and ending of the typical budgeting cycle, started to receive explicit attention. They were: (1) the way in which the budget is prepared (e.g. Hofstede, 1967; Milani, 1975); and (2) the use of budgets to control and evaluate managerial performance (e.g. Hopwood, 1972; Otley, 1978). Regarding budget preparation, studies by Hofstede and Milani suggested that the e€ects of budgeting depended on the extent to which subordinate managers were allowed to participate in the budgeting process and the extent to which they where capable of in¯uencing budget target levels. This line of research was continued by studies that focused on a construct formalized as budget participation (e.g. Ronen & Livingstone, 1975; Brownell, 1981; Brownell & Hirst, 1986; Dunk, 1989). An overview of the early budget participation literature is found in Brownell (1982a). A recent and more critical overview of this literature is given by Shields and Shields (1998). Regarding the use of the budget as a tool for performance evaluation, Hopwood (1972) and later Otley (1978) focused on the use of budgetary information by superiors to evaluate their subordinate's performance. As the latter two studies are commonly regarded as the ``formal'' starting point for RAPM research (e.g. Kren & Liao, 1988; Briers & Hirst, 1990), separate attention is devoted to them below.

453

2.1. Dysfunctional consequences of RAPM In a study by Hopwood (1972) the budgetpressure construct from Argyris (1952) and DeCoster and Fertakis (1968) was replaced with the more focused construct of supervisory style. Rather than following these two studies in their reliance on the budget-pressure construct (cf. Swieringa & Moncur, 1972), Hopwood (p. 157) started from the idea, supported by anecdotal evidence, that the extent and manner in which budgets are used reflect di€erent management styles. In particular, Hopwood questioned whether dysfunctional reactions to budgeting were the e€ect of the inherent characteristics of accounting performance measures that budgets contain, or the e€ect of the precise manner in which and extent to which superiors used these measures to evaluate the performance of their subordinate managers.1 Hopwood's supervisory style construct was developed to express this ``manner and extent''.2 Hopwood (pp. 157±158) pointed to four inherent limitations of accounting information for managerial performance assessment: First, not all the relevant dimensions of managerial performance are included in accounting reports since neither accountants nor managers have developed comprehensive measures and standards. Second, even considering the economic aspects of performance, an organization's economic cost function is rarely known with precision and an accounting system can only attempt to approximately represent its complexity (...). Third, the accounting data are primarily concerned with representing outcomes, while 1 Note that by focusing on the e€ects of performance evaluation, the budgeting theme became intertwined by a theme concerning the dysfunctional e€ects of performance evaluation, that had recently started to receive academic attention as well (e.g. Ridgway, 1956). 2 In later RAPM studies, the supervisory style construct has received alternative and synonymous labels like budget emphasis (e.g. Brownell & Dunk, 1991), supervisory evaluation style (e.g. Harrison, 1992) and RAPM (e.g. Hirst, 1981).

454

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

managerial activity is concerned with the detailed process giving rise to the ®nal outcomes (...). Fourth, the main emphasis of accounting reports is on short-term performance indexes while the evaluation of managerial performance is often concerned with more longterm considerations. The type of reasoning Hopwood (1972) applied in developing hypotheses was straightforward and corresponded with the role-theoretic reasoning underlying DeCoster and Fertakis' (1968) budgetpressure construct. The general expectation was that dysfunctional behavior associated with the use of accounting performance information resulted from superiors neglecting these defects and using budgets in an absolute or Budget-Constrained way, rather than in a more relaxed Pro®t-Conscious way.3 In particular, Hopwood (1972) expected that the strict use of budget standards to evaluate managerial performance would negatively a€ect managers' job-related tension, job attitudes, and social relations, and would indeed spur dysfunctional decision making and data manipulation. Such e€ects were argued to be the result of: ... disagreement and con¯ict between (the subordinate) and his supervisor over the dimensions, and their values, on which the job is described and evaluated. (Hopwood, 1972, p. 161). As the quotation illustrates, the particular theoretical expectation was that strict emphasis on budget attainment would cause role con¯ict, which was identi®ed previously as a major source of job-related tension (Kahn et al., 1964; Rizzo et al., 1970). The relevance of the concept is displayed in Kahn's et al. (p. 19) formal de®nition of role con¯ict, as: 3 Hopwood designed a measurement scale for this study, using the relative importance of eight performance criteria which received both a rank and a score. The measurement scale was based on interviews, containing performance evaluation related statements which were common language in the organization investigated (see, Hopwood, 1972, 1973).

... the simultaneous occurrence of two (or more) sets of pressures such that compliance with one would make more dicult compliance with the other.4 Hopwood (1972, p. 161) furthermore argued that too much budget emphasis would not only cause disagreement and con¯ict, but would also be ine€ective since: ... even if the cost-center manager tries to improve his performance in terms of the accounting indexes, the behavior which is necessary to achieve this is not always clear if some of the reported costs are not under his control. Hopwood's (1972) study focused on cost-center managers, and was based on interviews and a questionnaire survey. The analysis showed overall support for the expected e€ect of the supervisory styles outlined. The Hopwood study was replicated by Otley (1978) who studied the e€ects of RAPM in a pro®t-center environment. Otley's ®ndings essentially contradicted Hopwood's ®ndings, as he found no signi®cant relations between budget emphasis and either job-tension or negative social relations. Furthermore, Otley (1978) found positive relations between budget emphasis and managers' budgetary performance, which falsi®ed Hopwood's suggestion that a strict reliance on APM would be universally inappropriate. 2.2. Dysfunctional consequences of RAPM in context The contradictory ®ndings in the Hopwood (1972) and Otley (1978) studies provided a strong incentive for further empirical research.5 Many later studies explicitly refer to these two studies for motivating their own e€orts (e.g. Brownell, 1982b; Hirst, 1981, 1983b; Govindarajan, 1984). Several factors have contributed to the ``pioneering'' 4 The full argument concerning role con¯ict is presented in the more elaborate Hopwood (1973). 5 Notably Hopwood and Otley have stayed out of the ``debate'' ever since starting it.

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

status of these two studies. First, both studies contained a comprehensive analysis of the e€ects of supervisory style, using a variety of budget-related and work-related variables. Second, the studies were largely similar concerning research questions and research method, suggesting that the di€erence in results was, somehow, ``genuine''. Third, Otley (p. 123) himself triggered suspicion that the di€erent results could be caused by systematic di€erences between his and Hopwood's samples, by noting: The present study was designed to eliminate technical failings in the accounting system, as far as possible, by observing the operation of a well designed system in a type of organization that was well suited for the application of budgetary control. This notion of the suitability of RAPM laid the track for an explicit inclusion of contextual variables in research models that could explain di€erent ®ndings from situational di€erences between the Hopwood and Otley samples (Kren & Liao, 1988, p. 282).6 The common argument in these studies is that accounting reports only potentially have inherent defects, such as those outlined by Hopwood. However, the extent to which they are truly defective, and the extent to which their use truly results in dysfunctional e€ects, depends on the exact organizational context in which they are used. Therefore, while Hopwood is responsible for creating the distinction between budget system and its use to explain dysfunctional e€ects of budgeting, now context was added as a third, and theoretically relevant, dimension. Over the last two decades, several contextual variables have been investigated in an attempt to understand the conditions under which RAPM is more (or less) e€ective. Together, the e€orts form the area of the literature, called RAPM research, which has received the positive and negative quali6 The suitability of budgetary control for the sample studied was based on earlier studies (e.g., Bruns & Waterhouse, 1975), suggesting that decentralized and independent subunits (i.e. pro®t-centers) had the ``pro®le of the optimum situation for budgetary control'' (Otley, 1978, p. 126).

455

®cations mentioned at the outset of this paper. Common elements of RAPM studies are the focus on the use of accounting information for managerial performance evaluation, the frequent use of contingency frameworks in which the contextual appropriateness of RAPM is analyzed, and the heavy reliance on research methods from psychological and sociological research ®elds (cf. Merchant & Simons, 1986; Brownell, 1995). Table 1 presents an outline of some key aspects of the RAPM studies, organized by contingency variable examined. Below, attention is devoted to each category of studies, ranging from the ``broad'' national culture variable, to the more ``speci®c'' personality variables. A ®rst category deals with universal studies, of which the early Hopwood and Otley studies are also examples. Because many studies have used multiple contingency factors, and because many studies have tested multiple hypotheses, both the table and the discussion contain cross references. 2.3. Universal (non-contingency) studies Several universal studies have been done as well since the Hopwood (1972) and Otley (1978) studies. Onsi (1973) found, in line with Hopwood, that higher RAPM was associated with a greater propensity to created budgetary slack, and to engage in ``creative'' accounting, caused by a tendency to enhance the attainability of budget targets. In several later studies this ®nding has been (partially) replicated (Merchant, 1985c; Hughes & Kwon, 1990; Lal, Dunk, & Smith, 1996). A study by Kenis (1979) investigated the e€ects of RAPM on an array of job-related variables, such as job tension and job satisfaction. RAPM appeared to have a negative e€ect on tension, but, contrary to prediction, several positive e€ects of RAPM were found as well. These e€ects included an increase in managers' motivation to participate in the budgeting process. Similarly, Hirst and Yetton (1984) noted as a positive e€ect that RAPM decreased managers' role ambiguity. Overall, the results provide rather strong support for the e€ects of RAPM on slack creation and data manipulation. However, also positive e€ects of RAPM have been reported.

456

Table 1 Overview of RAPM studies Relevant hypothesis: RAPM more appropriate for ...

Criterion (dependent) variable(s)

Support for hypothesis (claimed)

Findings signi®cant

Sample (No. of organizations)1

Other hypotheses or variables in this study

Statistical model

A. None (universalistic) 1. Hopwood (1972) 167 supervisors (1)

(No contingency variable)

Job-related tension Relations with supervisor Relations with peers Data manipulation Dysfunctional behavior Pairwise t-test2

Yes Yes Yes Yes Yes

Partially Partially Partially Partially Partially

2. Onsi (1973) 132 functional mgrs (7)

(No contingency variable) exploratory

Budgetary slack Data manipulation Correlation

n.a. n.a.

Yes Yes

3. Otley (1978) 39 pro®t center mgrs (1)

(No contingency variable)

Job-related tension Trust in supervisor Evaluation clarity Job clarity Evaluation fairness Budgetary performance Correlation

No Yes Yes Yes No Yes

No Yes Yes Yes No Yes

4. Kenis (1979) 169 department heads (16)

(No contingency variables)

Job-involvement Job-satisfaction Job tension Budget attitudes Budget motivation Budgetary performance Cost efficiency Job performance Linear regression

No No Partially3 No No Partially Partially Partially

No No Partially Partially Yes Yes Partially Partially

5. Hirst and Yetton (1984) = Hirst (1983b), see C1

(No contingency variable)

Role ambiguity Linear regression

Yes

Yes

6. Merchant (1985c) = Merchant (1981), see E1

(No contingency variable)

Budgetary slack Correlation

Partially4

Partially

See also E4

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Contingency factor Study (year)

(No contingency variable)

8. Hughes and Kwon (1990) = Merchant (1981), see A5 9. Lal et al. (1996) 83 production mgrs (83) B. National culture characteristics 1. Harrison (1992) 211 sales/purchase mgrs (28) 2. Harrison (1993) = Harrison (1992), see B1

Accrual manipulation (3) Partially Short-term thinking Yes Discouragement new ideas Yes Correlation

Partially Yes Yes

(No contingency variable) exploratory

Budgetary slack LISREL

n.a.

Yes5

(No contingency variable)

Budgetary slack Correlation

Yes

Yes

Higher budgetary participation (regardless of nation) See also F8

Job-related tension Job satisfaction 3-Way interaction

Yes Yes

No No

Cultures with higher power-distance and collectivism (Singapore vs. Australia; nation = dummy)

Job-related tension Job-satisfaction 2-Way interaction Job-related tension Job-satisfaction Job-related tension Job-satisfaction 3-Way interaction

Yes Yes

Yes Yes

n.a. n.a. Not tested Not tested

Yes No ± ±

See also C6, D6 and G3

Exploratory: nation  personality (authoritarianism) Exploratory: nation  personality (collectivism) See also H2

3. Lau et al. (1995) 114 functional mgrs (80)

For hypotheses, see E11

No formal test

Not tested

±

C. Environmental characteristics 1. Hirst (1983b) 111 functional mgrs (N.R.)6

Medium (environmental and task) uncertainty

Job-related tension Relations w. supervisor Quadratic regression Job-related tension 2-Way interaction

No No

No No

Yes

Yes

Lower (environmental and task) uncertainty 2. Merchant (1984) = Merchant (1981), see E1

3. Govindarajan (1984) 58 SBU mgrs (18)

Products later in PLC Products with stronger market position

Use of APM Use of APM Subgroup z-test See also E3

No No

No No

Lower environmental uncertainty

Effectiveness 2-Way interaction

Yes

n.a.7 457

(Table continued overleaf )

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

7. Merchant (1990) = Merchant (1985b), see D2

458

Table 1Ðcontd from p. 456 Relevant hypothesis: RAPM more appropriate for ...

Criterion (dependent) variable(s)

Sample (No. of organizations)1

Other hypotheses or variables in this study

Statistical model

4. Brownell (1987b) 50 functional mgrs (1)

Lower environmental uncertainty

Job performance Job satisfaction 2-Way interaction

5. Ezzamel (1990) 81 mgrs (81)

Higher environmental uncertainty

Use of APM Correlation

6. Merchant (1990) = Merchant (1985b), see D2

Lower environmental uncertainty See also A7, D6 and G3

Support for hypothesis (claimed)

Findings signi®cant

Yes Yes

Yes Yes

Accrual manipulation (3) Correlation

Partially

Partially

7. Ross (1995) = Ross (1994), see G6

Lower (environmental and task) uncertainty

No

No

D. Strategic characteristics 1. Govindarajan and Gupta (1985) = Govindarajan (1984), see C38

Job-related tension ANOVA

Build strategies than harvest strategies

Performance 2-Way interaction

Partially

Partially

Exploratory: harvest vs. growth strategies Pro®t-centers performing worse

Use of accounting controls n.a. Use of accounting controls Partially Subgroup z-test

Yes Partially

3. Gupta (1987) = Govindarajan (1984), see C3

Build strategies than harvest strategies9

Performance 2-Way interaction

Yes

Yes

4. Simons (1987b) 76 mgrs (76)

Defender than prospector firms

Use of tight budget goals Use of cost control Logit regression

No Yes

No Yes

5. Govindarajan (1988) 121 general SBU mgrs (24)

Low cost than differentiation strategy

Effectiveness 2-Way interaction

Yes

Yes

2. Merchant (1985b) 54 pro®t center mgrs (1)

See also E8 and H2 6. Merchant (1990) = Merchant (1985b), see D2

Growth than maintain/harvest strategies See also A7, C6, and G3

Long-range orientation No No Discouragement new ideas No No Subgroup z-test

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Contingency factor Study (year)

E. Task and departmental characteristics 1. Merchant (1981) 170 manufacturing mgrs (19) 2. Hirst (1983b) 111 functional mgrs (N.R.) 3. Merchant (1984) = Merchant (1981), see E1

Larger, decentralized firms

Partially Partially

Partially n.a.10

Use of APM Correlation Performance Subgroup z-test Use of APM Correlation Performance Subgroup z-test

Partially

Partially

Partially

Partially

Yes

Yes

Yes

Yes

For hypotheses, see C1 Routine and repetitive production technology

see also C2 Production than non-production jobs Exploratory: low vs. high job structure

Role ambiguity Role ambiguity Linear regression

Yes n.a.

Yes Yes

Marketing than R&D managers

Performance 2-Way interaction

No

No

No Yes

No Yes

7. Macintosh and Daft (1987) 90 functional mgrs (20)

Higher budget participation (low task uncertainty subgroup) Performance Job-related tension 3-Way interaction Higher sequential interdependence

Yes

Partially

8. Govindarajan (1988) 121 SBU mgrs (24)

Use of operating budgets Correlation

Match: decentralization, strategy and personality

Performance Correlation11

Yes

Yes

Job stress Performance Budget attitudes 2-Way interaction Job stress Performance Budget attitudes 2-Way interaction

No No No

No No No

No No No

No No No

5. Brownell (1985) 66 R&D and marketing mgrs (1) 6. Brownell and Hirst (1986) 76 functional mgrs (1)

9. Imoisili (1989) 102 cost center mgrs (3)

See also D5 and H1 Lower task uncertainty

Lower task interdependence

459

(Table continued overleaf )

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Larger, more diverse departments

4. Hirst and Yetton (1984) = Hirst (1983b), see C1

Use of APM performance Subgroup z-test

460

Table 1Ðcontd from p. 458 Relevant hypothesis: RAPM more appropriate for ...

Criterion (dependent) variable(s)

Support for hypothesis (claimed)

Findings signi®cant

Sample (No. of organizations)1

Other hypotheses or variables in this study

Statistical model

10. Williams et al. (1990) 201 mgrs (22)

Exploratory: pooled vs. reciprocal task interdependence

Performance dimensions Canonical correlation

n.a.

Partially

11. Brownell and Dunk (1991) 79 functional mgrs (46)

Lower budget participation  task uncertainty

Performance 3-Way interaction Performance 2-Way interaction

Yes

Yes

Yes

Yes

Performance 2-Way interaction

Yes

Yes

Job-related tension 3-Way interaction Higher budget participation (low task uncertainty subgroup) Job-related tension 2-Way interaction Lower task difficulty  budgetary participation Performance 3-way interaction Higher budget participation (low task difficulty subgroup) Performance 2-Way interaction See also B3 and F9

No

No

Not tested

Not tested

Yes

Yes

Yes

Yes

Higher budget participation (low task difficulty subgroup) 12. Dunk (1992) = Dunk (1989), see F3 13. Ross (1995) = Ross (1994), see G6 14. Lau et al. (1995) 114 functional mgrs (80)

See also F6 Higher levels of manufacturing process automation For hypotheses, see C7 Lower task uncertainty  budgetary participation

15. Abernethy and Brownell (1997) 127 senior R&D ocers (2)

Tasks with lower analyzibility and fewer exceptions

Partially

Partially

F. Budget participation 1. Brownell (1982a, 1982b) 38 cost-center mgrs (1)

Performance 2-Way interaction

Higher budget participation

Job satisfaction Performance 2-Way interaction

No Yes

No Yes

Higher budget participation

Job performance 2-Way interaction

No

No

Exploratory: lower vs. higher budget participation

Performance 2-Way interaction

Yes

Yes

2. Brownell and Hirst (1986) 76 functional mgrs (1) 3. Hirst (1987b) 44 managers (1) 4. Dunk (1989) 26 production mgrs (26)

For hypotheses, see E5

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Contingency factor Study (year)

5. Aranya (1990) 97 store mgrs (97) 6. Brownell and Dunk (1991) 79 functional mgrs (46) 7. Dunk (1993) = Brownell and Dunk (1991), see E1113

G. Superior-subordinate relation factors 1. Hopwood (1974) = Hopwood (1972)14

2. Merchant (1985b) 54 pro®t center mgrs (1) 3. Merchant (1990) see A7 4. Macintosh and Williams (1992) = Williams et al. (1990), see E9 5. Dunk (1993) = Brownell and Dunk (1991), see E11 6. Ross (1994) 215 responsibility center mgrs (18) H. Personality characteristics 1. Govindarajan (1988) 121 general SBU mgrs (24) 2. Harrison (1993) = Harrison (1992), see B1

Job satisfaction Budgetary performance 2-Way interaction

n.a. n.a.

Yes Yes

Budget slack 3-Way interaction

No

Yes

Use of APM Use of APM Pairwise t-test

Yes Yes

Partially Partially

For hypotheses, see E11 Lower budget participation  information asymmetry For hypotheses, see B1 For hypotheses, see E14

Lower consideration Higher initiating structure See also I1 Exploratory: initiating structure leadership style Exploratory: considerate leadership style

Use of accounting controls No Use of accounting controls No Correlation

No No

Considerate leadership style

Accrual manipulation Subgroup z-test

No

No

Performance Canonical correlation

n.a.

Partially

Job-related tension ANOVA

Yes

Yes

Job-related tension Job satisfaction 2-Way interaction Job-related tension Job satisfaction 2-Way interaction

No No

Yes No

Not tested15 Not tested

Not tested Not tested

See also A7, C6, and D6 Managerial roles For hypotheses, see F6 Higher trust in supervisor

For hypotheses, see E8 Higher individual's authoritarianism Higher individual's collectivism (= lower individualism) See also B3

461

(Table continued overleaf )

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

8. Harrison (1992) 211 sales/purchase mgrs (28) 9. Lau et al. (1995) 114 functional mgrs (80)

Exploratory: lower vs. higher budget participation12

462

Table 1Ðcontd from p. 460 Contingency factor Study (year)

Relevant hypothesis: RAPM more appropriate for ...

Criterion (dependent) variable(s)

Sample (No. of organizations)1

Other hypotheses or variables in this study

Statistical model

I. Contagion e€ect 1. Hopwood (1974) = Hopwood (1972), see A1

Exploratory: managers evaluated by APM themselves

1

See also G1 Managers evaluated by APM themselves

Findings signi®cant

Use of APM Not statistically tested

Partially

Not tested

Use of APM Correlation

Yes

Yes

When the sample has been used in an earlier study, it is mentioned after the sign `='. When it is noted `see X', it means that the study is the same as study X. The results reported here concern Hopwood's (1972) general hypothesis. This hypothesis was tested using pairwise t-tests for di€erent means between subgroups based on the three evaluation styles (Budget Constrained, Pro®t Conscious and Non-Accounting). Note that these three subgroups were based on the relative importance of `Meeting the Budget' and `Concern with Cost' among eight performance criteria. Other and later `exploratory' analyses in this study concerned calculating correlation coecients for scores on the two individual performance criteria mentioned and the dependent variables. Here many opposite results were reported, for example indicating that `Meeting the Budget' and `Concern with Cost' were negatively related to job-related tension. 3 Kenis (1979) tested both the e€ects of a general use of APM and a punitive use of APM. The expected negative e€ects of RAPM on the dependent variables were only partially found and, moreover, only for punitive use of APM. No negative e€ects, and even positive e€ects, were found for general use of APM. Overall, this is denoted here as `partial' support for the hypothesis. 4 RAPM was measured with three variables. `Budgetary slack' was signi®cantly and positively related to `reactions to expected budget overruns' and signi®cantly and negatively related to `required explanatioons of variance' and `link with extrinsic rewards'. 5 This study reports a reanalysis of Merchant's (1985c) data, which were used previously in Merchant (1981), using a LISREL-methodology. Now a positive relationship between RAPM and budgetary slack was found. 6 The acronym `NR' indicates `Not reported'. 7 Govindarajan (1984) compares correlation statistics for the relationship between environmental uncertainty and RAPM for more and less e€ective business-units. No formal test was reported. Furthermore, the relevant test should have concerned the e€ect of environmental uncertainty on the relationship between RAPM and e€ectiveness (see also text). 8 From the Govindarajan (1984) sample now 46 observations were used. The number of organizations was not reported in this paper. 9 See also Govindarajan (1984), here reported as D1. 10 The test used here, comparison of correlation coecients per subgroup using z-tests, is not well suited to test hypotheses predicting interaction e€ects. 11 Govindarajan (1988) calculated distance scores as the actual scores minus the optimal score for each of the independent variables. These distance scores were correlated with performance. This approach supports a systems concept of contingency ®t. 12 Aranya (1990) claims to have sampled in a high task uncertainty environment, and argues that for high task uncertainty, budgetary participation has a negative e€ect on the appropriateness of RAPM. 13 Number of organizations not reported. 14 Number of subjects not reported. 15 No formal tests were done after the initial analysis revealed that `individualism-collectivism' scores were not reliable. 16 Only 26 observations used. 2

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

2. Barrett et al. (1992) 72 marketing mgrs (15)

Support for hypothesis (claimed)

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

2.4. National culture characteristics RAPM studies paying attention to national culture examine whether budget-based performance evaluation is more appropriate in some countries, than in others. The number of studies in this category is limited. Harrison (1993) found, in a setting of Australian and Singaporean managers, a direct e€ect of national culture on the relationship between RAPM and attitudinal variables. In Singapore, where power-distance is high and individualism is low, RAPM had a less negative e€ect on job-related tension and a more positive e€ect on job satisfaction. In the same setting, Harrison (1992) had earlier found no e€ect of national culture on the interaction between budgetary participation and RAPM. Also a study by Lau et al. (1995) was motivated by the aspect of national culture. They investigated, similarly to Harrison (1992), whether previous ®ndings concerning budgetary participation were transferrable to Singapore (i.e. to a non-Anglo-Saxon country). The e€ect of national culture was not formally tested, but the results did not suggest a great impact. Overall, these studies provide at best partial support for the relevance of national culture. 2.5. Environmental characteristics A large group of studies has tried to explain di€erences in the appropriateness of RAPM from the organization's external environment. Typically, these studies have focused on factors contributing to the variability, unpredictability or uncertainty of the environment. Hirst (1983b) attempted to solve the Hopwood±Otley controversy by pointing out that RAPM was the least appropriate when uncertainty was either high or low. With high uncertainty, APM are relatively incomplete, causing superior±subordinate con¯icts, and job-related tension. Alternatively, low uncertainty would cause APM to be ``too'' complete, resulting in a loss of subordinates' discretionary power, and in con¯ict and job-related tension as well. Hirst's analysis revealed, contrary to the expected curvilinear e€ect, that the appropriateness of RAPM, in terms of job-related tension, was low (high) when uncertainty was high

463

(low). Regarding superior±subordinate relationships, no signi®cant e€ects were found. In a study by Govindarajan (1984) similar hypotheses were tested in a sample of business-unit managers. Govindarajan predicted that APM were used less when environmental uncertainty was high, and that this uncertainty±RAPM relationship would be more ``pronounced'' for more e€ective businessunits. The analysis showed a negative correlation between uncertainty and use of APM, which was more negative for more e€ective business units.7 Later studies investigating the same or similar hypotheses showed mixed results. In a direct attempt to replicate Hirst's ®ndings, Ross (1995) failed to ®nd a negative e€ect of uncertainty on the appropriateness of APM. Furthermore, Ezzamel (1990) found a positive correlation between uncertainty and use of APM. In sum, it appears that the overall evidence regarding environmental uncertainty is mixed at best. 2.6. Strategic characteristics A third set of contingency studies has investigated the appropriateness of RAPM for di€erent strategies. The literature on strategy-control system relationships, of which these studies are a subset, has recently been reviewed by Lang®eldSmith (1997). Overall, she ®nds that the number of strategy-control studies is limited, and she concludes that strong evidence is lacking. An important cause may be that di€erent typologies of strategy are used, leading to incomparability of ®ndings (Lang®eld-Smith, 1997, p. 209). Also for RAPM studies investigating strategy, the results are mixed. Govindarajan and Gupta (1985) expected that RAPM would be less appropriate for business-units following a build strategy than for business-units following a harvest strategy, but found no support. APM appeared to be equally e€ective for build and harvest strategies, 7 Govindarajan (1984) did not test whether the di€erence in correlations between high performers and low performers was statistically signi®cant. Furthermore, the proper test for ®nding whether RAPM should be matched with environmental uncertainty would have meant that the relationship between RAPM and performance would be di€erent for di€erent levels of environmental uncertainty.

464

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

even if in ``build'' business units the use of supplementary, long-term, performance criteria positively a€ected performance. Govindarajan (1988) found that RAPM is more appropriate for business-units following a low-cost strategy instead of a di€erentiation strategy, but other studies show no or only partial support (Merchant, 1985b, 1990). Simons (1987b) found that the use of ®nancial controls di€ered between defender and prospector ®rms. The use of tight budget controls appeared, however, to be associated with higher performance for both categories of ®rms. These results and the di€erences in strategic typologies used, give little room for strong conclusions. 2.7. Task and departmental characteristics The largest group of RAPM studies tries to establish a link between characteristics of the subordinate's task and department, and the appropriateness of RAPM for his or her evaluation. The theoretical arguments underlying hypotheses for the often studied task uncertainty variable are largely similar to those related to environmental uncertainty. The Hirst (1983b) paper has already been mentioned. The instrument used to measure uncertainty in that study also contained elements related to task uncertainty. An unsuccessful attempt to replicate Hirst's results was made by Brownell and Hirst (1986). This study combined an analysis of task uncertainty with budget participation (see below). Later, Brownell and Dunk (1991) found support for an e€ect of task variability (and not task diculty) on the e€ect of RAPM, a ®nding that was partially replicated by Lau et al. (1995). Also in these latter two studies, the task uncertainty variables and budget participation variables were examined simultaneously, hindering the interpretation of ®ndings for task uncertainty alone (cf. Southwood, 1978). Other studies focused on broader (organizational) variables than managerial tasks alone, suggesting that APM are more appropriate for larger, formally controlled departments (e.g. Merchant, 1981, 1984). In sum, the evidence for this category of studies suggests that the appro-

priateness of RAPM tends to be higher for larger, more interdependent departments, and less uncertain tasks. 2.8. Budget participation A large class of studies investigates the combined e€ects of two budget-related factors, RAPM and budgetary participation. Budgetary participation research investigates the (dys)functional consequences of subordinate participation in setting budget levels. Functional consequences may result from information sharing between superior and subordinate, and from a positive e€ect on subordinates' goal-acceptance and motivation (e.g. Brownell & McInnes, 1986; Dunk, 1993). Dysfunctional consequences may result from subordinates' attempts to negotiate slack into their budgets (e.g. Dunk). Brownell (1982a) found that budget participation and RAPM should be matched for optimal performance in the sense that when participation is high (low), RAPM should also be high (low). This ®nding has been combined with the ®ndings of Hirst (1983b) in several subsequent studies. The combined and more elaborate hypothesis was that the match between budgetary participation and RAPM only holds in low task uncertainty situations (e.g. Brownell & Hirst, 1986; Brownell & Dunk, 1991; Lau et al., 1995). Overall, the ®ndings of these studies are mixed. Furthermore, in many studies the results are not easily interpretable, because of the combined analysis of budget participation and task uncertainty. In a simpler research model Dunk (1989) and Aranya (1990), moreover, found e€ects contradicting Brownell's (1982b) original ®nding. In sum, it appears that the combined e€ects of RAPM and budgetary participation are not yet well understood. 2.9. Superior±subordinate relation factors and the contagion e€ects A broad category of RAPM studies encompasses contingency variables that describe aspects of the social relationship between superior and subordinate. Both Hopwood (1974) and Merchant

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

(1985b, 1990) studied the relationship between leadership style and RAPM. Hopwood found some support for the expectation that RAPM is matched with more considerate and less initiating structure leaders, but these ®ndings could not be replicated by Merchant (1985a, b, c, 1990). Ross (1994) studied trust, arguing that this variable reduces the potential e€ect of RAPM on role con¯ict. He found that RAPM is more appropriate when managers trust their supervisors. Based on the principal-agent framework, Dunk (1993) investigated the e€ect of information asymmetry between superior and subordinate. His results, although opposite to expectations, suggested that it may be a relevant variable. Finally, Hopwood and Barrett, McDonagh, and Granleese (1992) showed that superiors tend to match their evaluation style with the style used in their own performance evaluation (see section I in Table 1). This phenomenon is known as the contagion e€ect (cf. Ansari, 1977, p. 110). Overall, the evidence suggests that factors concerning superior-subordinate relationships may be important, but so far only limited evidence exists. 2.10. Personality factors A ®nal and small category of contingency variables relates to the personality of the subordinate manager. Govindarajan (1988) found that in the optimal situation, strategy was matched with the subordinate's locus of control and RAPM. In particular, the expectation and ®nding was that the optimal combination for a di€erentiation strategy would be an external locus of control, and low RAPM. Harrison (1993) found no support for the expectation that RAPM was more appropriate, in terms of job-related tension, for people with higher authoritarianism. The expectation was that based on the assumption that people with higher authoritarianism, would more easily accept the use of APM in hierarchical settings. In contrast, the results supported an opposite relationship. None of the relationships with job satisfaction were statistically signi®cant. Also for this last category of RAPM studies, the overall results do not support strong conclusions. In the next section reasons will be presented for this state of a€airs.

465

3. RAPM research: an evaluation The overview of RAPM papers above supports the idea of a critical mass, at least in terms of research volume. Furthermore, many RAPM studies explicitly attempt to build on ®ndings of previous studies, and attempt to integrate and replicate previous ®ndings (cf. Lau et al., 1995, p. 360). However, and despite the common elements in RAPM studies outlined before, the idea of an organized critical mass of RAPM studies seems to be challenged by both methodological and theoretical problems. Important symptoms are not only the relatively large number of hypotheses not supported, but even more the low success rate of studies aimed at explicit replication of previous ®ndings. Concerning the latter, recall that Brownell and Hirst (1986) and Ross (1995) tried but failed to replicate the Hirst (1983b) ®ndings. Hirst (1987b), Dunk (1989) and Aranya (1990) tried but failed to replicate Brownell (1982b). Other replication studies were only partially successful (e.g. Brownell & Dunk, 1991). Consequently, Lindsay and Ehrenberg (1993, p. 224) even mention the RAPM paradigm as an example of an area of unsuccessful replications, and note: Taken as a whole, this body of research, although typically `interesting' in seeking to explain discrepancies, does not add up to a coherent body of knowledge or understanding. In a recent overview of survey studies in management accounting, which included also many RAPM studies, Young (1996, p. 55) makes a similar remark, as he notes that: ... 25 years of survey research has yet to yield a cohesive body of knowledge about management accounting and control practices within and across industries and nations. These comments, in particular, provide a sharp contrast with the idea of an ``organized critical mass'' of RAPM studies. The following three subsections discuss in more detail the problems of RAPM research that may explain these expressions of severe critique. The ®rst subsection below

466

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

discusses problems concerning the meaning and measurement of the RAPM construct. The second subsection addresses limitations in the basic formulation of RAPM theory. The third subsection discusses problems with contingency theory in RAPM research. 3.1. Meaning and measurement of the RAPM construct Despite the central role of the RAPM construct, RAPM studies show no unity in its empirical measurement. Brownell and Dunk (1991, p. 702) even note: ...the measurement of budget emphasis in performance evaluation (RAPM) is far from settled. That researchers have achieved as much towards reconciling the Hopwood (1972) and Otley (1978) results is quite remarkable when one considers the variations of measurement of budget emphasis. Many studies use Hopwood's (1972) original instrument, but formats di€er from the original in the number and wording of the criteria (e.g. Otley, 1978; Brownell & Hirst, 1986), and in the way of deriving scores for statistical analysis (e.g. Brownell, 1982b; Harrison, 1992, 1993). A second class of studies use the BPQ, deriving RAPM scores through factor analysis (e.g. Kenis, 1979; Merchant, 1981, 1984), which has resulted in different numbers of factors, and di€erent factor contents across studies. A third class of studies use ad-hoc ``RAPM'' instruments (e.g. Hirst, 1983b; 8 Hirst (1983b) developed and used an instrument that measures the reliance on quantitative performance information. Govindarajan (1984) and Govindarajan and Gupta (1985) measured RAPM as the extent to which managerial bonuses depend on objective performance data (``formula-based approach'') as opposed to superiors' subjective considerations (``subjective approach''). Both studies explicitly point out that the instrument they used was designed to be equivalent to the Hopwood instrument. For example Govindarajan (p. 125) states: ``It is worth noting that strictly `formula-based' and strictly `subjective' approaches are analogous to Hopwood's `budget-constrained' and `non-accounting' styles. In addition, a combination of subjective and formula-based approaches is similar to Hopwood's `pro®t-conscious' style.''

Govindarajan, 1984).8 In none of these studies is formal proof provided for the equality or validity of the instruments used. Moreover, individual studies do not seem to acknowledge the importance of consistent measurement across studies in a developing paradigm (cf. Lindsay & Ehrenberg, 1993). Although variable measurement is typically considered a methodological issue, the variety in RAPM measurement also has important theoretical implications. First, the variety hinders the de®nition of the boundaries of the RAPM paradigm. For example, Govindarajan explicitly aims to solve the Hopwood±Otley controversy. In later studies, that use the same data and the same ``RAPM'' measures (e.g., Govindarajan & Gupta, 1985; Gupta & Govindarajan, 1984), this link is less explicitly made. Therefore, it is not clear to which exact phenomena ``RAPM knowledge'' pertains. Second, the diversity in RAPM measurement raises the question about the essence of the RAPM construct, and even if the RAPM construct still means `the same' after several decades of research (e.g., Ross, 1995). In most studies, RAPM conveys a notion of targets expressed in accounting numbers, which therefore are of a quantitative and ®nancial nature (e.g. Hofstede, 1967, p. 26; Hirst, 1987a; Harrison, 1993, p. 319).9 The emphasis on the accounting nature of APM causes them to be associated with rigidity, formality and objectivity (Lang®eld-Smith, 1997; Chapman, 1997), and their use is subsequently associated with job-related tension and disagreement. In other studies, RAPM re¯ects the use of budgets, and APM are regarded as typical examples of formal controls (cf. Fisher, 1995). While APM are clearly related to both, the concepts of targets and formal controls are not equivalent. Consequently, predictions about the appropriateness of RAPM will di€er between the concept chosen, and the same applies for de®nitions of the ``opposite'' of 9

Ross (1995, p. 3) argues that the content of these measures is di€erent: ``In each of the ®ve questions used by Hirst [1983a, pp. 603±604] the emphasis is on the use of quantitative criteria for performance evaluation. (...) Although ®nancial information is part of the quantitative information that could be used for performance evaluation it is also likely that non-®nancial quantitative information would be included in a manager's de®nition of `quantitative'.''

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

RAPM. If RAPM is presented as a formal control, its e€ects should be contrasted with the use of informal controls (e.g. self-controls). If RAPM is presented as the use of accounting targets, its e€ects should be compared with those of ``other'' targets (non-®nancial, subjective, long-term, qualitative etc.). Interestingly, already Kahn (1972) questioned the theoretical focus on ``the budget'' in the Hopwood (1972) study, and the underdevelopment of the (opposite) Non-Accounting style, but so far no formal analysis has been done.10 Finally, the variety in RAPM measurement and the underdevelopment of the RAPM concept have probably also hindered its foundation in the conceptual management control literature. Neither the RAPM construct itself, nor any of the styles conceived by Hopwood have found their direct theoretical counterpart in the management control typologies outlined by Merchant (1982, 1985a), Ouchi (1977, 1979), Anthony (1981), Anthony and Govindarajan (1995) and others. This may also have reinforced the vague and imprecise equation of RAPM with ``formal'', ``hierarchical'' and ``static'' controls (cf. Chapman, 1997).11 3.2. A theory of RAPM A second set of problems relates to the basic formulation of RAPM theory. In their overview of early RAPM literature, Briers and Hirst (1990, p. 385) sharply criticize the underdevelopment of theory in many RAPM studies: Of particular concern is the inclusion of variables in hypotheses with little supporting 10 Notably, already Hopwood's original study exempli®es the importance of strict RAPM-de®nition. In this study, the RAPM styles in the statistical analysis were the result from preceding interviews that were done to validate the styles. In addition, Hopwood used more than one measure for RAPM. While this study is known for the discrete supervisory styles (and the negative e€ect of the Budget Constrained style), many of the negative e€ects were not found in analyses using the continuous accounting-related variables. 11 RAPM would typically be classi®ed as a `bureaucratic' control in Ouchi's framework (1977, 1979). Its classi®cation in Merchant's (1982, 1985a) framework would be ambiguous, as it contains aspects of both `action control' and `results control'.

467

explanation. For example, some studies use box diagrams with arrows indicating causally related variables. Although this is a parsimonious way of communicating connections, the supporting argument in some studies is only suggestive (...). Briers' and Hirst's (1990) remark is general, without a precise speci®cation of ``good'' and ``bad'' studies. Furthermore, their remark seems to apply more to the presentation of theory in the papers than to the use of theory itself. In fact, and in contrast with the diversity in measurement of the RAPM construct, the paradigm shows great consistency in its basic theoretical framework, that is informed by role theory.12 This relative unity especially re¯ects in the criterion variables used to test RAPM hypotheses. Role theory typically points to such interpersonal and work-related factors as job-related tension and job satisfaction. Moreover, the emphasis on role theory (with its focal concept of role con¯ict) seems the cause of the limited focus in RAPM studies on the potential negative e€ects of RAPM, which speci®cally re¯ects in the frequent use of job-related tension. However, many studies have used performance as a criterion variable, rightly arguing that it is the `ultimate' variable of interest. Unfortunately, the use of performance-related variables is problematic for other reasons (cf. Smith, 1983; Briers & Hirst; Lang®eld-Smith, 1997). First, RAPM research is about ®nding the right performance measures, and it seems illogical that a researcher uses the right performance measures, when the whole paradigm is about ®nding ``right'' performance measures. Indeed, Young (1996) points out that survey instruments for measuring performance too easily presuppose ``relevant'' dimensions of performance. Second, there is reason to suggest that performance is an independent variable (i.e. an antecedent of RAPM), rather than a 12 Notable exceptions to the role-theoretic framework are the use of balance theory (Brownell, 1982b), and the of principal-agent models (Dunk, 1993). Yet, Brownell's alternative theory is severely critiqued by Briers and Hirst (1990), and the reference to principal-agent theory by Dunk merely results in the adoption of new variables (i.e. information-asymmetry), not a new theory.

468

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

dependent variable (e.g. Merchant, 1985b; Lang®eld-Smith, p. 226). Finally, studies often display a gap between the theoretical hypothesis and the empirical test, by developing predictions about, for example, job-related tension and subsequently testing performance. Yet, it has often been demonstrated that the relationships between job attitudes (e.g. tension, satisfaction) and performance are complex and unclear (e.g. Schuler, 1980; Jamal, 1984; Tosi, Rizzo, & Carroll, 1994). Furthermore, such gaps obscure whether a failure to ®nd support for a hypothesis is caused by insucient theory or faulty method. Recently, Shields and Shields (1998, p. 50) derive the same conclusion regarding budgetary participation research. In sum, RAPM studies appear to be limited in their use of theory, and in their choice of criterion variables for testing the appropriateness of RAPM. Although this apparent unity could, by itself, positively a€ect the structure of the paradigm, the disappointing ®ndings suggest that the paradigm would bene®t from a more speci®c de®nition, measurement and modeling of appropriateness. 3.3. A contingency theory of RAPM Briers' and Hirst's (1990) comment on the underdevelopment of RAPM theory is particularly troublesome, since it echoes many of Otley's comments made ten years before, in a review of early contingency studies in management accounting. Otley (1980, p. 414) noted: The contingency approach is invoked, so it seems, in order to cover up some of the embarrassing ambiguities that exist in the universalistic approach. Contingency theory is often associated with the adage ``it all depends'', but Otley's comment seems to indicate that the adage signals the absence rather than presence of theory. RAPM studies show a great diversity in applying contingency frameworks, which forms a third source of problems hindering the organization and integration of the paradigm. A ®rst source of diversity is the fact that RAPM studies vary widely in level of analysis, with contingency variables ranging from the broad concept of national culture to much

more focused personality variables. Consequently, evidence is limited for each individual contingency variable. A second source of diversity is the di€erence between studies in the motivation to include a certain contextual variable. In many studies, the contextual variable chosen is argued to account for a systematic situational di€erence between the Hopwood and Otley samples (Kren & Liao, 1988, p. 282). Such studies, for example, explicitly investigate alleged di€erences between cost-center (Hopwood) and pro®t-center (Otley) managers (e.g. Hirst, 1983b). Other studies however, although also motivated by the contradictory ®ndings from these two early studies, use fewer theoretical or systematical considerations in choosing contextual factors. They focus on ad-hoc factors expected to in¯uence relationships examined earlier. Examples of the latter type of studies are those that combined the earlier mentioned budget-participation construct, to explain situational di€erences in the appropriateness of RAPM (e.g. Brownell, 1982b; Dunk, 1989). Third, a variety of motives exists to use contingency frameworks. In most cases, the motive is pragmatic, which is shown by using merely the outlines of contingency theory to study ``any'' contextual variable.13 Far less important seems the aim to contribute to develop a contingency theory of accounting (cf. Chapman, 1997), or to extend `the' contingency theory of organizations to management accounting (cf. Schoonhoven, 1981). A cause may be that RAPM studies seem typical examples of management theory, which aims to solve practical problems rather than explain accounting phenomena from the context in which they appear (cf. Puxty & Chua, 1989, p. 115).14 A 13 In RAPM research, for example, personality quali®es as a contingency variable, although it is not a ``contingency factor'' in a traditional sense (e.g., environment, strategy and organizational structure). 14 Puxty and Chua (1989, p. 115) note a di€erence between organization theory and management theory: ``Organisation theorists (...) ask questions concerning the way organisations operate. Management theorists take as a much closer focus the question of what a manager should do to make his organisation operate better''. This distinction also underlies the debate about ``normative'' versus ``positive'' accounting (e.g. Christenson, 1983). Lindsay (1995) notes that current RAPM studies have little to o€er in terms of advice to management (p. 44, 47).

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

relevant fourth source of variance is that RAPM studies carelessly mix predictions about the e€ects of APM with predictions about the use of APM. Some studies use a selection approach to contingency ®t (cf. Drazin & Van de Ven, 1985; Fisher, 1995; Selto, Renner, & Young, 1995). They predict that, under certain circumstances, superiors will be reluctant to use APM and will use other performance measures instead (e.g., Merchant, 1984; Govindarajan, 1984). Other studies use an interaction approach to contingency ®t, predicting certain (dys)functional e€ects of the interaction between RAPM and a contextual variable (Imoisili, 1989; Govindarajan, 1988; Brownell & Dunk, 1991). In many cases, no theoretical reason is given for the approach selected, and sometimes, the two kinds of expectations are even combined. For example, Merchant predicts an e€ect of technology and department size on both the use of APM (selection) and the e€ects of RAPM (interaction).15 Despite these di€erences, RAPM studies show a remarkable limited set of statistical models for testing contingency hypotheses. The dominant format in testing RAPM contingency hypotheses is the use of moderated regression analysis using interaction terms, in which the appropriateness of RAPM is the outcome of a RAPM-contingency interaction. Although this method is well documented in the methodological literature (e.g. Southwood, 1978; Cohen & Cohen, 1983; Champoux & Peters, 1987; Jaccard, Turrisi, & Wan, 1990), its use in RAPM research often does not seem the outcome of a deliberate choice, but rather seems the result of mere conservatism. Its unquestioned use limits the extent to which the RAPM paradigm can pro®t from developments in the wider organizational behavior literature, which documents many alternative formats for the formulation and statistical analysis of contingency hypotheses (see, e.g. Venkatraman & Camillus, 1984; Venkatraman, 15 Indeed, the ``systems approach'', the third alternative to express contingency ®t, was already propagated by Otley (1980) but has been used sporadically. See Govindarajan (1988), Abernathy and Brownell (1997) for exceptions. See Selto et al. (1995) for a (non-RAPM) combination of the three forms of ®t in one study.

469

1989).16 A ®nal and methodological point of critique is the typical use of small and non-random samples (cf. Lindsay, 1995). Sample sizes below 50 are typical rather than exceptional. This itself does not ``forbid'' the use of statistical analysis and statistical signi®cance, but it seems that signi®cance is often but wrongly equated with materiality, and chance of replication (cf. Lindsay). The particular reason for mentioning this problem in this subsection on contingency theory, is that the statistically demanding interaction models used to test contingency hypotheses provide very unstable results with these low sample sizes (Schmidt & Hunter, 1978; Cohen & Cohen). This is a likely, additional, cause for the lack of consistent and replicated ®ndings across studies (cf. Lindsay & Ehrenberg, 1993). 4. RAPM and uncertainty The overview and analysis of RAPM research above could easily lead to pessimistic opinions about its current state, and its future. Many reviews of management accounting (and RAPM) research have indeed been quite successful in presenting gloomy pictures (cf. Chapman, 1997). However, to conclude the discussion in such a negative spirit, would be both pointless and shortsighted. An important reason for optimism is that many of the methodological problems noted can be resolved. For example, Young (1996) notes that in itself, survey research is a respectable way for data collection, and the notion of contingency theory is never challenged. Lindsay (1995) argues that a way to resolve current ¯aws in methodology would be to use more random sampling, and to focus more on simple and controlled replication (cf. Lindsay & Ehrenberg, 1993). Indeed, these and other aspects of the typical methodology of RAPM research are well documented in the literature (e.g., Dillman, 1978; Kerlinger, 1986). 16 A di€erent although related issue concerns the use and interpretation of interaction statistics. Hartmann and Moers (1998) have recently shown that the application and interpretation of interaction statistics in RAPM research is subject to frequent and serious errors.

470

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

A greater cause for concern, and a greater challenge for research, are the problems related to RAPM theory. They will be the focus of the remaining sections of this paper. While methodological ¯aws may a€ect individual studies, theoretical ¯aws a€ect a whole research paradigm. Moreover, it is possible that the disappointing results from RAPM studies so far may be explained by a new theoretical perspective. Recently such a new approach for studying the role of accounting systems in organizations was proposed by Chapman (1997). Chapman (p. 189) points to the clarity and precision of seminal contingency research in the sixties (e.g. Burns & Stalker, 1961; Woodward, 1965; Lawrence & Lorsch, 1967), and points to the sharp contrast with the current state of contingency research in management accounting. He speci®cally proposes that contingency frameworks in management accounting should focus on uncertainty as the central concept, and he uses Galbraith's (1973, 1977) framework for ``organizational design'' to support his arguments. Galbraith de®nes uncertainty as an information de®cit, and his framework provides a synthesis of early contingency frameworks in which uncertainty also was the central concept (Chapman, p. 199). In the remainder of this paper, an attempt is made to transfer and extend Chapman's proposal to the RAPM literature. In the subsections below, ®rst an overview is given of the current `status' of uncertainty in the RAPM literature. Based on this overview it will be concluded that uncertainty provides the RAPM paradigm with a paradox. The section closes by discussing how the explicit recognition of the paradox may prove bene®cial for the further organization and development of the RAPM paradigm. 4.1. The focus on uncertainty in RAPM studies Apart from the opposite ®ndings of Hopwood (1972) and Otley (1978), another important factor spurring the in¯uence of contingency theory in RAPM research was the prior development of the contingency theory of organizations, and the resulting availability of ``ready made'' contingency frameworks to management accounting researchers

(Otley, 1980, p. 416; Kren & Liao, 1988, p. 283). In the seventies, several contingency frameworks for management accounting and control were developed, which were either normative and directed at prescribing accounting and control system design (e.g. Lorange & Scott Morton, 1974; Gordon & Miller, 1976; Waterhouse & Tiessen, 1978; Amigioni, 1978; Daft & Macintosh, 1978), or which served to test hypotheses predicting ``®t'' between selected organizational characteristics and the accounting system (e.g. Bruns & Waterhouse, 1975; Hayes, 1977). These studies had inconclusive ®ndings (Sathe, 1978, p. 91; Hopwood, 1978, p. 8; Otley, 1980, p. 417), but they served to illustrate how contextual factors typical for contingency research in organizational behavior could be used in management accounting research, which are the uncertainty associated with the organization's environment and with its task technology (e.g. Burns & Stalker, 1961; Woodward, 1965; Lawrence & Lorsch, 1967; Thompson, 1967; Perrow, 1967, 1970; cf. Sathe, 1978, p. 183; Otley, p. 419). In these early studies, the organizational environment was captured in terms of its dynamism, heterogeneity (Gordon & Miller, 1976), predictability (Waterhouse & Tiessen, 1978), complexity and variability (Amigioni). Task technology in turn was described as the variety (Daft & Macintosh) and routineness (Waterhouse & Tiessen) of tasks. The focus on uncertainty re¯ected the prime importance of this concept in the organizational behavior literature. Regarding organizational design, Thompson (1967, p. 159) argued that uncertainty is the ``fundamental problem'' of complex organizations and that dealing with uncertainty is the `essence' of administrative processes. Similarly, Galbraith (1977, p. 36) had noted that: Uncertainty is the core concept upon which the organization design frameworks are based. The importance is re¯ected in empirical research as well, as evidenced by Downey and Slocum's (1975, p. 562) argument that uncertainty had become the ``central concept'' of organizational contingency research. Similarly, Miles and Snow (1978, p. 254) concluded that:

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

... the concept of uncertainty has emerged as a primary variable linking a great number of organizational characteristics to conditions in the environment. 4.2. RAPM and uncertainty: theory With this heritage, it is not surprising that uncertainty was among the ®rst variables in RAPM studies adopting a contingency perspective (Hirst, 1983b; Govindarajan, 1984; cf. Brownell, 1987b; Chapman, 1997). It is often not recognized, however, that in other and later RAPM studies, uncertainty has remained an important factor, often serving as the implicit or theoretical factor behind the explicit and operational contingency variable chosen for empirical analysis. For example, Merchant (1984) investigated di€erences in the appropriateness of RAPM for subordinate managers of departments operating with di€erent production technologies and in di€erent markets, based on the expected di€erence in the uncertainty that the two kinds of departments would face. Brownell (1985) investigated di€erences in the appropriateness of RAPM between R&D and production departments, based on the assumption that the two would face di€erent levels of environmental uncertainty. Both Imoisili (1989) and Macintosh and Daft (1987) investigated departmental interdependence and argued that this latter variable was an important cause of uncertainty in the manager's working environment. Also the much studied budget participation variable is uncertainty related, as it in¯uences information asymmetry between subordinate and superior (e.g. Dunk, 1993), and is often a response to task uncertainty and environmental uncertainty (cf. Shields & Shields, 1998). In RAPM hypotheses, uncertainty is typically predicted to a€ect subordinates' perceptions of such factors as the controllability, completeness and relevance of RAPM. According to these arguments, uncertainty manifests itself as changing conditions that a€ect the controlled process. Uncertainty causes predictions to be dicult and thus hinders budgetary target setting. In addition, uncertainty makes it dicult to judge ex-post whether good (poor) managerial performance measured in accounting terms is the

471

result of good (poor) managerial e€ort or (un)lucky circumstances (e.g. Govindarajan, 1984, p. 128; Brownell, 1985; Williams, Macintosh, & Moore, 1990; Imoisili, 1989). In short, dysfunctional consequences of using RAPM under uncertainty result from not applying or not being able to apply the controllability principle, which forms a central element of responsibility accounting frameworks (see below). Stressing `uncontrollable' APM can result in the incompleteness of performance measurement (e.g., Hayes, 1977; Hirst, 1981, p. 771; cf. Govindarajan & Gupta, 1985; Merchant, 1990; Ross, 1995), or even in the irrelevance of performance evaluations (Govindarajan, p. 128; Govindarajan & Gupta, p. 54). The typical (role-theoretic) hypothesis predicts that uncertainty negatively a€ects the appropriateness of RAPM. 4.3. RAPM and uncertainty: evidence Although the role of uncertainty in extant RAPM research is easily demonstrated, the current state of RAPM research does not allow the conclusion that the e€ects of uncertainty are well understood. An important example is the Hirst (1981, 1983b) study. Recall that Hirst originally expected a curvilinear e€ect of uncertainty on the appropriateness of RAPM. So far, no empirical proof has supported this initial expectation, but the linear relationship Hirst found instead has been wrongly equated with his original theory ever since (see, e.g. Brownell, 1987a). Yet, even this ®nding can, for several reasons, not be generalized to an overall negative relationship between uncertainty and the appropriateness of RAPM. First, in later studies that attempted to replicate Hirst's ®ndings, no interaction e€ects of uncertainty and RAPM on job-related tension were found (Brownell & Hirst, 1986; Imoisili, 1989; Brownell & Dunk, 1991; Lau et al., 1995; Ross, 1995). Second, Hirst did not ®nd support for the expected interaction e€ect of RAPM and uncertainty on superior-subordinate relations, which he proposed as an alternative measure for the appropriateness of RAPM, following Hopwood (1972). This at least suggests that conclusions about the contextual appropriateness of RAPM depend on the

472

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

speci®c criterion variable used. Third, Hirst's ®nding that RAPM is less appropriate under higher levels of uncertainty have not generally been con®rmed by studies focusing on the performance e€ects of using RAPM under uncertainty. Merchant (1984), Brownell (1985) and Lau et al. did not ®nd evidence of uncertainty a€ecting the relationship between the use of RAPM and managerial performance, which they predicted. Since, in contrast, studies by Brownell (1987a) and Govindarajan (1988) con®rm a negative e€ect of uncertainty on the relationship between RAPM and managerial performance, the results are mixed at best. Fourth, also studies predicting an e€ect of uncertainty on the use of APM by superiors provided mixed results. Merchant only found partial support for an e€ect of uncertainty on the use of RAPM. Govindarajan (1984) found a negative relationship between environmental uncertainty and RAPM, but did not test whether the use of APM was negatively related to performance under high uncertainty (see before). Govindarajan and Gupta (1985) found that non-accounting information was used more in SBU's with high-uncertainty strategies, but they also found that RAPM was equally e€ective in both high and low-uncertainty strategies. Finally, many studies suggest and ®nd that RAPM is especially useful in situations of high uncertainty (cf. Simons, 1987a, p. 341). Ezzamel (1990) found that APM were used more frequently under conditions of high uncertainty, the rationale of which was that larger, decentralized ®rms would emphasize formal communication means when faced with uncertainty. Ezzamel's study builds upon and con®rms earlier ®ndings (e.g., Khandwalla, 1972; Merchant, 1981; Simons, 1987b) that a positive relationship exists between the uncertainty originating in market factors (e.g. competition) and the use of formal controls such as budgets and accounting controls (cf. Chapman, 1997). Similarly, Macintosh and Daft (1987) found a positive relationship between departmental interdependence, an important source of uncertainty (e.g., Thompson, 1967) and the emphasis placed on meeting budgetary targets. In sum, therefore, the results concerning RAPM and uncertainty do not display a large amount of consistency. In fact, the support for hypotheses

predicting both a positive e€ect and a negative e€ect provides us with an apparent paradox. 4.4. The uncertainty paradox The overview presented above suggests that no support exists for a universal negative e€ect of uncertainty on the appropriateness of RAPM. This clearly falsi®es claims that the results of RAPM studies relying on uncertainty are consistent (e.g., Brownell, 1987a; Chapman, 1997, p. 193). In particular, such claims suggest that the results of a positive e€ect of uncertainty on the appropriateness of RAPM are neglected (cf. Briers & Hirst, 1990).17 Poole and Van de Ven (1989, pp. 562±563) explain this negligence by pointing to the general aim of a researcher in a speci®c domain to signal consistency (cf. Gresov, 1989; Mak, 1989). They note: Contemporary theory construction methods are biased toward consistency. Relatively little attention has been paid to the resolution of tensions or oppositions. and: The presence of contrary of contradictory assumptions, explanations, or conclusions is often seen as an indicator of poor theory building In contrast, Poole and Van de Ven (1989, p. 563) argue that explicit recognition of a paradox is a fruitful exercise in the progress of theories, and recommend that more attention is given to theoretical tensions and inconsistencies, to stimulate the development of more comprehensive theories. Claims for the consistency of ®ndings concerning uncertainty in RAPM research indeed fail to recognize that uncertainty provides us with a paradox for the design of management control systems. Recall, that in traditional RAPM thinking, and in the design of management control 17 Briers and Hirst (1990) note many examples of ``selective referencing'' in RAPM research, although they do not explicitly point out these results.

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

systems, the controllability principle is paramount. This principle states that: ... the manager of a responsibility center should be assigned responsibility only for the revenues, costs, or investment that responsibility center personnel control (Atkinson, Banker, Kaplan, & Young, 1997, p. 564). The principle seems particulary applicable to uncertain situations, where the risk of being held responsible for uncontrollable factors is high. On the one hand, it seems widely accepted that adherence to the principle is desirable, since it represents that performance evaluations should be `fair' (e.g. Merchant, 1987, p. 317). Choudhury (1986, p. 189) thus notes: ... the notion of being answerable (only) for what one is able to in¯uence, may be viewed as conforming to a commonly held concept of justice ... On the other hand, it should be noted that a strict application of the principle would imply that in most cases the use of budgets for managerial performance evaluation (and management control) would not be feasible, since especially in uncertain and complex organizations, with people working in joint e€ort, the possibility to single out individual responsibilities is severely limited (cf. Hirst, 1983a, p. 29; Merchant, 1987). The tension between the apparent logic of the controllability principle, and its apparent limited practical applicability has lead to di€erent opinions about its value. Hirst (p. 36) explicitly concludes that the principle is paradoxical. On the one hand, decentralization is an important means of enhancing the controllability of organizational processes, and budgets are important tools to formalize decentralization, by means of creating responsibility centers (cf. Bruns & Waterhouse, 1975; Merchant, 1981, 1984). On the other hand, a strict application of the controllability principle would hinder the subsequent use of the budget for managerial performance evaluation. This and similar paradoxes, which point to the apparent limited feasibility of formal organizational controls in cases where

473

they may be most needed, are mentioned throughout the control literature (e.g. Ouchi & Maguire, 1975; Vancil, 1979). Ouchi and Maguire (1975, p. 568) point to the paradox of using output performance measures with decentralization, stating: The use of output measures is largely a result of the demand for quanti®able, simple measures. Paradoxically, output measures are used most when they are least appropriate: in the face of complexity (and) interdependence. Baker, Jensen, & Murphy (1988, p. 598) note a tension between decentralization as a way to pro®t from ``speci®c knowledge'' of the subordinate on the one hand, and the subsequent obligation for the superior to choose the ``correct objective measure of subordinate performance'' on the other hand. Vancil (1979) addresses as a general theme the `ambiguity' associated with decentralization and control, noting that control implies freedom and restriction simultaneously. Thompson (1967, p. 145) points to the ``paradox of administration'', noting that organizations' administrations aim at ¯exibility, while being mainly occupied with the reduction of uncertainty. Merchant (1984) points out that planning and control systems cannot be ¯exible and in¯exible simultaneously. Finally, Emmanuel, Otley and Merchant (1990, p. 183) point out that budgets are most feasible, where they are least needed and vice versa. In sum, the discussion above suggests support for the relevance and importance of uncertainty when explaining and predicting the appropriateness of RAPM. However, it is also likely that the lack of consistent ®ndings in RAPM studies thus far is the consequence of the tension between the opposite e€ects of using APM under uncertainty. In the ®nal section some suggestions are given to deal with this tension. 5. Implications for RAPM research According to Poole and Van de Ven (1989), the reasons to expose paradoxes lie in the alleged bene®ts for theoretical progress. In this case, several ways exist to address the uncertainty paradox

474

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

of RAPM research, and each of these ways may prove to be fruitful for the further development of the RAPM paradigm. In this ®nal section, attention will be given to ®ve speci®c directions for further RAPM research that share the aim to resolve the paradox, and to revive and further organize the RAPM paradigm. A ®rst such direction would involve an explicit investigation of the di€erent roles that APM play in evaluative situations, since it is not clear that APM takes the same role in each organization or division (cf. Kren & Liao, 1988, p. 300). Recently, Chapman (1997, p. 201) suggested that the role of accounting systems in general may depend on the level of uncertainty. Typically, accounting systems function as ``answer machines'' in low-uncertainty situations, but they may become `learning machines' when uncertainty increases. Simons' (1995) recent control framework, may also support such multiple roles for accounting information and budgets. Whereas accounting budgets typically serve as `diagnostic controls' (feedback systems) under relative certainty, they also provide `boundaries' to management activity, and may function as `interactive controls', when operational targets should be aligned with strategy and (strategic) uncertainty (Simons, p. 138; Horngren, Sundem, & Stratton, 1996, p. 257). Notably, already Simon, Kozometsky, Guetzkow and Tyndall's (1954) study emphasized that management accounting information is collected for various purposes; problem-solving, attention directing and scorecard keeping activities. This understanding of the multiple roles of accounting and of RAPM may be enhanced if researchers explicitly consider which essential aspect of RAPM they address. An interesting illustration is provided by Govindarajan (1984, p. 128) who argued that APM are not appropriate under high uncertainty, and noted: Performance evaluation presupposes targetsÐ either explicit (e.g. budgets) or implicit. To arrive at a priori targets that can serve as valid standards for subsequent performance evaluation, one must be able to predict the conditions that will exist during the coming year.

Further research should be able to explain why uncertainty would only harm the controllability of accounting targets, which Govindarajan (1984) does not, and thus whether the essence of APM under uncertainty lies in the target-nature or in the accounting-nature of APM. Since it has been found that (formal) strategic control systems can be rigid as well (Ittner & Larcker, 1997, p. 294), another relevant dimension of APM under uncertainty to be examined may be its status of a formal control system (e.g. Fisher, 1995). Second, research should become more explicit about the meaning of appropriateness. So far, RAPM studies have generally argued that uncertainty limits the feasibility of using RAPM, concluding that RAPM will be more appropriate under conditions of low uncertainty. Although this bias toward dysfunctional side-e€ects can be explained by looking at the history of the paradigm (see, e.g. Argyris, 1952), and although understanding dysfunctional e€ects should be considered important, the resulting analyses are often only partial. Instead, the appropriateness of RAPM should be approached as a combination of functional and dysfunctional consequences. In other words, while the feasibility of controls (and RAPM) is an important element in control system design (e.g. Merchant, 1985a), so is the need for control (e.g. Merchant, 1982, 1985a). This importance of the need for controls is currently often forgotten, possibly by the dominant focus on role con¯ict. A framework that may further substantiate the need for controls, and that may explain the positive e€ects of RAPM under uncertainty, could be based on goal theory (cf. Kren & Liao, 1988, p. 289). Typical goal-theoretic variables are ``goal clarity'' and `goal speci®city', and the goal-theoretic literature advocates, and ®nds, that clear and speci®c goals and objectives lead to higher performance (McConkie, 1979; Locke, 1968; Locke & Latham, 1990; Tosi et al., 1994). This may explain the positive e€ects of RAPM, since APM are relatively clear and speci®c (cf. Tosi, 1975, p. 150; Schuler, Beutell, & Youngblood, 1989, p. 240), as the results from Hirst and Yetton (1984) seem to support. It would be speci®cally important to study the e€ects of goal-setting under uncertainty. Hirst (1987a) pro-

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

vides an initial theoretical analysis of the goalsetting e€ects of budgets under uncertainty, but so-far no empirical work has been done in this ®eld. Further RAPM research may, therefore, build on studies in organizational behavior (e.g., Early, 1985; Wood, Mento, & Locke, 1987; Wood, Bandura, & Bailey, 1990) that examine the e€ects of goal-setting in various contexts. An alternative framework for analyzing the e€ects of RAPM could be based on `equity theory'. Equity theory is concerned with peoples' perceptions of ``fairness'' and ``justice'' in social relationships (cf. Huseman, Hat®eld, & Miles, 1987; Landy, 1989), and has been used to predict and explain perceptions about the fairness of performance evaluations and rewards in organizational settings (Landy, Barnes, & Murphy, 1978; Landy & Farr, 1980; Landy, Barnes-Farrell, & Cleveland, 1980). Recall, that ``fairness'' has been mentioned often as an important reason for subordinates resistance against `uncontrollable' performance measures (see, e.g. Hopwood, 1972, p. 161; Otley, 1978, p. 132; Lau et al., 1995, p. 360). The relevance of this variable for RAPM research is further demonstrated by Ross (1994). He found that trust between subordinate and superior mitigated the job-related tension e€ects of using APM. Trust is positively a€ected by the perceived `fairness' of performance evaluations (see, e.g. Fulk, Brief, & Barr, 1985).18 Although some of these alternative frameworks may be more fruitful than others, they at least serve to demonstrate the possibilities for a more explicit de®nition of the concept of appropriateness, and to demonstrate possible alternatives to role theory in developing predictions about the contextual appropriateness of RAPM. In any case, their use will often imply the use of other criterion variables than those currently used in RAPM studies, that are more explicitly chosen, and that provide a better link between theory and test (cf. Shields & Shields, 1998, p. 50). 18 Magner, Welker and Campbell (1995) found that the trust-in-supervisor is, reversely, a€ected by the perceived fairness of budgetary targets. Whatever the exact causal direction of the relationship of ``fairness'' and ``trust'', they do support the relevance of the concept of `fairness'. Lindquist (1995) shows that `fairness' in¯uenced the acceptability of budgetary goals.

475

Third, theory is needed about the appropriateness of RAPM under uncertainty for the performance evaluator. In RAPM studies `appropriateness' is examined in terms of subordinate's attitudes and responses. An alternative level of analysis would be the superior, and a related question would be whether superiors' evaluative behaviors can be explained in terms of contextual `appropriateness'. Especially studies investigating the contagion e€ect (e.g. Hopwood, 1974; Barrett et al., 1992) could be extended to explain why superiors have a tendency to pass performance criteria on to lower-level managers, and whether this especially happens with accounting performance measures. At least part of the explanation might lie in the uncertainty superiors face in observing and evaluating subordinates' behaviors. Keeley (1977) found that supervisors may reduce their own uncertainty by emphasizing objective performance measures, such as APM, which limit supervisors' need for subjective judgment (cf. Hartmann, 1995). Further studies in this direction may easily connect to studies in which RAPM is the dependent variable (e.g. Merchant, 1981, 1984). However, RAPM would now be modeled as the explicit, theoretically predicted, outcome, rather than as an indication of contextual ®t, based on a ``loose'' application of contingency thinking. Such more comprehensive contingency frameworks may also lead to a better understanding of the diversity of roles that APM play in the (control) relationship between superiors and subordinates. In particular, they would allow an explicit integration of attempts to explain the use of APM better, and attempts to provide practical recommendations for using APM, since they would more broadly consider the options and trade-o€s superiors face in choosing performance measures (e.g. Argyris, 1988; Puxty & Chua, 1989). These attempts could eventually result in a theory of superior choice, which is not available now. Some APM studies suggest that other factors than appropriateness in terms of subordinates' reactions determine superiors' choice of performance criteria. Otley (1978, p. 138) argues that actual performance in¯uenced the reliance on APM, suggesting that APM are emphasized more in units with low pro®tability (cf. Merchant, 1985b;

476

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Imoisili, 1989; Briers & Hirst, 1990; Lang®eldSmith, 1997), and the organizational behavior literature points to the importance of sociological and psychological variables that determine the choice of performance criteria (e.g. Rahman & McGosh, 1976; Neu, 1992; Judge & Ferris, 1993). Such a theory of superiors' choice would be the behavioral ``counterpart'' of principalagent models that do explicitly consider the use of performance evaluation criteria as the outcome of a superior's choice under uncertainty (e.g. Prendergast & Topel, 1993; Eisenhardt, 1985).19 Fourth, RAPM research should be challenged by ®nding out whether the appropriateness of RAPM is di€erently a€ected by di€erent kinds of uncertainty. Typically, the arguments supporting uncertainty-related expectations have not discriminated between types and sources of uncertainty. Potentially relevant di€erences include di€erences between complexity and uncertainty (cf. Khandwalla, 1972; Govindarajan, 1984; Ezzamel, 1990; Chapman, 1997) and between objective ``uncertainty'' and perceived uncertainty (Downey & Slocum, 1975; Downey, Hellriegel, & Slocum, 1975). In RAPM research, these nuances have generally not been recognized. In particular, advances can be made by giving theoretical meaning to di€erences between environmental uncertainty and task uncertainty. Especially the role of budgets as a bu€er against external uncertainty (cf. Thompson, 1967; Hayes, 1977; Merchant, 1984), and as a trigger to plan under environmental uncertainty (cf. Merchant, 1989, Chapter 5; Simons, 1987a, 1995) seem to deserve further investigations. In particular, these illunderstood functions of budgets under uncertainty may also explain a more positive role for budgets and RAPM under environmental uncertainty than is now generally hypothesized in studies that `mix' the e€ects of task uncertainty and environmental uncertainty. Finally, the impact of uncertainty on the appropriateness of APM could be better under19 Such a theory would also respond to the comment by Flamholtz and Das (1985) that the management control literature should be more explicit about the organizational level it addresses.

stood by including the personality of the subordinate manager. Current RAPM studies exclusively focus on uncertainty as a factor external to the manager, but there is reason to suggest that managers' personal and general attitudes toward uncertainty a€ect the appropriateness of RAPM. Personality factors have been mentioned before as important determinants of managerial behavioral and attitudinal reactions to budgeting (e.g. Murray, 1990, p. 119). Likely candidates for investigation are personality variables related to individual preferences for risk and uncertainty (cf. Duncan, 1972, p. 314; Chapman, 1997, p. 199). Especially a construct called tolerance for ambiguity may prove useful for this purpose (e.g. Budner, 1962; Norton, 1975). Tolerance for ambiguity expresses an individual's demand for information in uncertain environments (Frenkel-Brunswik, 1949; Martin & Westie, 1959; MacDonald, 1970) and seems, therefore, closely related to the conceptualization of uncertainty as a `de®cit in information' (cf. Galbraith, 1973, 1977). In related research ®elds (e.g. auditing, strategic management) this construct has shown to explain personal attitudes, behaviors and information preferences under uncertainty (e.g., Downey & Slocum, 1975; Duncan, 1972; Dermer, 1973; Gupta & Govindarajan, 1984; Gul, 1986). This variable may also explain a need for APM under uncertainty, since the clarity and precision of ®nancial and quantitative performance measures may reduce subordinates' perceptions of ambiguity (e.g., Gupta & Govindarajan; Hirst & Yetton, 1984). 6. Concluding comments The recommendations for further elaboration of the RAPM paradigm given above are based on the presumption that it is worthwhile to continue to direct academic attention to the budgeting phenomenon in organizations in general, and to the extent and manner in which budgets are used for controlling individual managerial behavior in particular. This presumption is based on the idea that using ``some form'' of accounting target setting will remain important, because, as Chapman (1997, p. 202) notes:

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

In a business setting (...) pro®t will still be the primary goal to be attained and so accounting cannot simply be discarded. Similarly, Parker and Lewis (1995, p. 212) note that the emphasis in management control on ``traditional'' management accounting techniques is persistent, and may even be ``topical'' because in business and management: ... the dominant themes of the 1990s have become ®nancial discipline, eciency measurement, market solutions and the commercial `bottom line' (...) This contemporary environment and its attendant philosophies have sponsored the persistence of the classical management approach towards control in both organizational management and management accounting. Furthermore, claims for the continuing importance of accounting seem widely supported by the empirical literature that evidences the important and steady role of various forms of budgeting in contemporary organizations (see, e.g. Umapathy, 1987; Skinner, 1990; Merchant & Riccaboni, 1990; Otley, 1990). But even when acknowledging the importance of management accounting traditions, current theoretical and practical developments in management control should be considered a challenge rather than a threat for `traditional' management accounting research. The RAPM literature may indeed provide an initial framework for analyzing recent ideas about the use of, for example, balanced scorecards and shareholder value for managerial performance evaluation (e.g. Otley; Kaplan & Norton, 1996). Based on RAPM research speci®c hypotheses could be formulated about the usefulness of having multiple performance indicators (as in balanced scorecards), about the e€ect of having non-®nancial or nonquantitative performance targets, or about the controllability problems associated with the use of performance measures derived from capital markets (as in shareholder value). So far, both theory development and empirical evidence concerning the (contextual) appropriateness of these innovative ideas in management control are limited.

477

Finally, it should be noted that the current status and lack of organization of the RAPM research paradigm should be no reason for discouragement or abandonment. RAPM studies address important and interesting phenomena, which are embedded in a rich environmental, organizational, departmental and individual context (cf. Flamholtz & Das, 1985). Such ``richness'' obviously hinders quick answers to quick research questions. This means, however, that not only proposals for further theoretical development, but also pleas for methodological improvements deserve serious support. In particular, there seems to be room for more, small, high-quality and theory-driven replication studies (cf. Lindsay & Ehrenberg, 1993). Recall that RAPM research is well grounded in a well-documented research methodology for data gathering and data analysis (e.g. Dillman, 1978; Kerlinger, 1986; Lindsay, 1995; Young, 1996). Certainly, these relative strengths should be considered when judging claims for the need for more subjective and casebased studies in management accounting. References Abernethy, M.S., & Brownell, P. (1997). Management control systems in research and development organizations: the role of accounting, behavior and personnel controls. Accounting, Organizations and Society, 22, 233±248. Amigioni, F. (1978). Planning management control systems. Journal of Business Finance and Accounting 5, 279±291. Ansari, S.L. (1977). An integrated approach to control system design. Accounting, Organizations and Society, 2, 101±112. Anthony, R.N. (1981). The management control function. Boston: Harvard Business School Press. Anthony, R.N., & Govindarajan, V. (1995). Management control systems. Chicago: Irwin. Aranya, N. (1990). Budget instrumentality, participation and organizational e€ectiveness. Journal of Management Accounting Research, 2, 67±77. Argyris, C. (1952). The impact of budgets on people. Ithaca, New York: The Controllership Foundation. Argyris, C. (1988). Producing knowledge that is generalizable and usable for practice: a review. Accounting, Organizations and Society, 13, 101±106. Atkinson, A.A., Banker R.J., Kaplan R.S., & Young S.M. (1997). Management accounting. Upper Saddle River: Prentice-Hall. Baker, G.P., Jensen, M.C., & Murphy, K.J. (1988). Compensation and incentives: practice vs theory. The Journal of Finance, 43, 593±616.

478

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Barrett, T.F., McDonagh, P., & Granleese, J. (1992). The contagion e€ect in performance evaluation: evidence from the evaluations of marketing managers: a research note. British Accounting Review, 24, 331±341. Birnberg, J.G., Turopolec, L., & Young, S.M. (1983). The organizational context of accounting. Accounting, Organizations and Society, 8, 111±129. Briers, M., & Hirst, M.K. (1990). The role of budgetary information in performance evaluation. Accounting, Organizations and Society, 15, 373±398. Brownell, P. (1981). Participation in budgeting, locus of control and organizational e€ectiveness. The Accounting Review, 56, 844±860. Brownell, P. (1982). Participation in the budgeting process: when it works and when it doesn't. Journal of Accounting Literature, 1, 124±153. Brownell, P. (1982). The role of accounting data in performance evaluation, budgetary participation and organizational e€ectiveness. Journal of Accounting Research, 20, 12±27. Brownell, P. (1985). Budgetary systems and the control of functionally di€erentiated organizational activities. Journal of Accounting Research, 23, 502±512. Brownell, P. (1987). The role of accounting information, environment and management control in multinational organisations. Accounting and Finance, 27, 1±16. Brownell, P. (1987b). The use of accounting information in management control. In K.R. Ferris & J.L. Livingstone, (Eds.), Management planning and control. Columbus Ohio. Brownell, P. (1995). Research methods in management accounting. Coopers and Lybrand, Accounting Research Methodology, Monograph No. 2, Melbourne. Brownell, P., & Dunk, A.S. (1991). Task uncertainty and its interactions with budgetary participation and budget emphasis: some methodological issues and empirical investigation. Accounting, Organizations and Society, 16, 693±703. Brownell, P., & Hirst, M.K. (1986). Reliance on accounting information, budgetary participation, and task uncertainty: tests of a three-way interaction. Journal of Accounting Research, 24, 241±249. Brownell, P., & McInness, M. (1986). Budgetary participation, motivation and managerial performance. The Accounting Review, 61, 587±600. Bruns, W.J., & Waterhouse, J.H. (1975). Budgetary control and organization structure. Journal of Accounting Research, 13, 177±203. Budner, S. (1962). Intolerance of ambiguity as a personality variable. Journal of Personality, 30, 29±50. Burns, T., & Stalker G.M. (1961). The management of innovation, London: Tavistock Institute. Champoux, J.E., & Peters, W.S. (1987). Form, e€ect size and power in moderated regression analysis. Journal of Occupational Psychology, 60, 243±255. Chapman, C.S. (1997). Re¯ections on a contingent view of accounting. Accounting, Organizations and Society, 22, 189± 205.

Choudhurry, N. (1986). Responsibility accounting and controllability. Accounting and Business Research, 16, 189±198. Christenson, C. (1983). The methodology of positive accounting. The Accounting Review, 58, 1±22. Cohen J., & Cohen P. (1983). Applied multiple regression/ correlation analysis for the behavioral sciences. Hillsdale: Erlbaum. Daft, R.L., & Macintosh, N.B. (1978). A new approach to design and use of management information. California Management Review, 21, 82±92. DeCoster, D.T., & Fertakis, J.P. (1968). Budget-induced pressure and its relationship to supervisory behavior. Journal of Accounting Research, 6, 237±246. Dermer, J.D. (1973). Cognitive characteristics and the perceived importance of information. The Accounting Review, 48, 511±519. Dillman, D.A. (1978). Mail and telephone surveys, the total design method. New York: John Wiley and Sons. Downey, H.K., & Slocum, J.W. (1975). Uncertainty: measures, research and sources of variation. Academy of Management Journal, 18, 562±578. Downey, H.K., Hellriegel, D., & Slocum, J.W. (1975). Environmental uncertainty: the construct and its application. Administrative Science Quarterly, 20, 613±629. Drazin, R., & Van de Ven, A.H. (1985). Alternative forms of ®t in contingency theory. Administrative Science Quarterly, 30, 514±539. Duncan, R.B. (1972). Characteristics of organizational environments and perceived environmental uncertainty. Administrative Science Quarterly, 17, 313±327. Dunk, A.S. (1989). Budget emphasis, budgetary participation and managerial performance: a note. Accounting, Organizations and Society, 14, 321±324. Dunk, A.S. (1992). The e€ects of managerial level on the relationship between budgetary participation and job satisfaction. British Accounting Review, 24, 207±218. Dunk, A.S. (1993). The e€ect of budget emphasis and information asymmetry on the relation between budgetary participation and slack. The Accounting Review, 68, 400±410. Early, P.C. (1985). In¯uence of information, choice, and task complexity upon goal acceptance, performance, and personal goals. Journal of Applied Psychology, 70, 481±491. Eisenhardt, K.M. (1985). Control: organizational and economic approaches. Management Science, C31, 134±149. Emmanuel, C., Otley, D., & Merchant, K.A. (1990). Accounting for management control. New York: Chapman and Hall. Ezzamel, M. (1990). The impact of environmental uncertainty, managerial autonomy and size on budget characteristics. Management Accounting Research, 1, 181±197. Fisher, J. (1995). Contingency-based research on management control systems: categorization by levels of complexity. Journal of Accounting Literature, 14, 24±53. Flamholtz, E.G., & Das, T.K. (1985). Toward an integrative framework of organizational control. Accounting, Organizations and Society, 10, 35±50.

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482 Frenkel-Brunswik, E. (1949). Intolerance of ambiguity as an emotional and perceptual personality variable. Journal of Personality, 18, 108±143. Fulk, J., Brief, A.P., & Barr, S.H. (1985). Trust-in-supervisor and perceived fairness and accuracy of performance evaluations. Journal of Business Research, 13, 301±313. Galbraith, J.R. (1973). Designing complex organizations. Addison-Wesley: New-York. Galbraith, J.R. (1977). Organization design, Reading: AddisonWesley. Gordon, L.A., & Miller, D. (1976). A contingency framework for the design of accounting information systems. Accounting, Organizations and Society, 1, 59±69. Govindarajan, V. (1984). Appropriateness of accounting data in performance evaluation: an empirical investigation of environmental uncertainty as an intervening variable. Accounting, Organizations and Society, 9, 125±135. Govindarajan, V. (1988). A contingency approach to strategy implementation at the business-unit level: integrating administrative mechanisms with strategy. Academy of Management Journal, 31, 828±853. Govindarajan, V., & Gupta, A.K. (1985). Linking control systems to business unit strategy: impact on performance. Accounting, Organizations and Society, 10, 51±66. Gresov, C. (1989). Exploring ®t and mis®t with multiple contingencies. Administrative Science Quarterly, 34, 431±453. Gul, F.A. (1986). Tolerance for ambiguity, auditors' opinions and their e€ects on decision making. Accounting and Business Research, 16, 99±105. Gupta, A.K. (1987). SBU Strategies, Corporate-SBU Relations, and SBU E€ectiveness in Strategy Implementation. Academy of Managerial Journal, 30, 477±500. Gupta, A.K., & Govindarajan, V. (1984). Business unit strategy, managerial characteristics and business unit e€ectiveness at strategy implementation. Academy of Management Journal, 27, 25±41. Harrison, G.L. (1992). The cross-cultural generalizability of the relation between participation, budget emphasis and jobrelated attitudes. Accounting, Organizations and Society, 17, 1±15. Harrison, G.L. (1993). Reliance on accounting performance measures in superior evaluative style, the in¯uence of national culture and personality. Accounting, Organizations and Society, 18, 319±339. Hartmann, F.G.H. (1995). Accounting information, MBO systems and performance judgment accuracy: a research note. Unpublished working paper, Maastricht University. Hartmann, F.G.H., & Moers, F. (1998). Statistical interaction in management accounting research. Paper presented at the 21st EAA Annual Conference, Antwerp. Hayes, D.C. (1977). The contingency theory of managerial accounting. The Accounting Review, 52, 22±39. Hirst, M.K. (1981). Accounting information and the evaluation of subordinate performance: a situational approach. The Accounting Review, 56, 771±784. Hirst, M.K. (1983). The controllability of ®nancial outcomes. Abacus, 19, 29±38.

479

Hirst, M.K. (1983). Reliance on accounting performance measures, task uncertainty and dysfunctional behavior: some extensions. Journal of Accounting Research, 21, 596±605. Hirst, M.K. (1987). The e€ects of setting budget goals and task uncertainty on performance: a theoretical analysis. The Accounting Review, 62, 774±784. Hirst, M.K. (1987). Some further evidence on the e€ects of budget use and budget participation on managerial performance. Australian Journal of Management, 12, 49±56. Hirst, M.K., & Yetton, P. (1984). In¯uence of reliance on accounting performance measures and job structure on role ambiguity for production and non-production jobs. Australian Journal of Management, 9, 53±62. Hofstede, G. (1967). The game of budget control. Assen: Van Gorcum. Hopwood, A.G. (1972). An empirical study of the role of accounting data in performance evaluation. Journal of Accounting Research, 10, 156±182. Hopwood, A.G. (1973). An accounting system and managerial behaviour. Westmead: Saxon House. Hopwood, A.G. (1974). Leadership climate and the use of accounting data in performance evaluation. The Accounting Review, 49, 485±495. Hopwood, A.G. (1978). Towards an organizational perspective for the study of accounting and information systems. Accounting, Organizations and Society, 3, 3±13. Horngren C.T., Sundem, G.L., & Stratton, W.O. (1996). Introduction to management accounting. Upper Saddle River: Prentice-Hall. House, R.J., & Rizzo, J.R. (1972). Role con¯ict and ambiguity as critical variables in a model of organizational behavior. Organizational Behavior and Human Performance, 7, 467± 505. Hughes, M.A., & Kwon, S.Y. (1990). An integrative framework for theory construction and testing. Accounting, Organizations and Society, 15, 179±191. Huseman, R.C., Hat®eld, J.D., & Miles, E.W. (1987). A new perspective on equity theory: the equity sensitivity construct. Academy of Management Review, 12, 222±234. Imoisili, O.A. (1989). The role of budget data in the evaluation of managerial performance. Accounting, Organizations and Society, 14, 325±335. Ittner, C.D., & Larcker, D.F. (1997). Quality strategy, strategic control systems, and organizational performance. Accounting, Organizations and Society, 22, 293±314. Jaccard, J., Turrisi, R., & Wan, C.K. (1990). Interaction e€ects in multiple regression. Newbury Park: Sage. Jamal, M. (1984). Job stress and job performance controversy: an empirical assessment. Organizational Behavior and Human Performance, 33, 1±21. Judge, T.A., & Ferris, G.R. (1993). Social context of performance evaluation decisions. Academy of Management Journal, 36, 80±105. Kahn, R.L. (1972). Discussion of: an empirical study of the role of accounting data in performance evaluation. Journal of Accounting Research, 10, 183±186.

480

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Kahn, R.L., Wolfe, D.M., Quinn, R.P., & Snoek, J.D. (1964). Organizational stress: studies in role con¯ict and ambiguity. New York: John Wiley and Sons. Kaplan, R.S., & Norton, D.P. (1996). The balanced scorecard. Boston: Harvard Business School Press. Keeley, M. (1977). Subjective performance evaluation and person-role con¯ict under conditions of uncertainty. Academy of Management Journal, 20, 301±314. Kenis, I. (1979). E€ects of budgetary goals characteristics on managerial attitudes and performance. The Accounting Review, 54, 707±721. Kerlinger, F.N. (1986). Foundations of behavioral research. Fort Worth: Holt, Rinehart and Winston. Khandwalla, P.N. (1972). The e€ect of di€erent types of competition on the use of management controls. Journal of Accounting Research, 10, 275±285. Kren, L., & Liao, W.M. (1988). The role of accounting information in the control of organizations: a review of the evidence. Journal of Accounting Literature, 7, 280±309. Lal, M., Dunk, A.S., & Smith, G.D. (1996). The proprensity of managers to create budgetary slack: a cross-national reexamination using random sampling. The International Journal of Accounting, 31, 483±496. Lang®eld-Smith, K. (1997). Management control systems and strategy: a critical review. Accounting, Organizations and Society, 22, 207±232. Landy, F.J., Barnes, J.L., & Murphy, K.R. (1978). Correlates of perceived fairness and accuracy of performance evaluation. Journal of Applied Psychology, 63, 751±754. Landy, F.J., & Farr, J.L. (1980). Performance rating. Psychological Bulletin, 87, 72±107. Landy, F.J. (1989). Psychology and work behavior. Paci®c Grove: Brooks/Cole. Landy, F.J., Barnes-Farrell, J., & Cleveland, J.N. (1980). Perceived fairness and accuracy of performance evaluation: a follow-up. Journal of Applied Psychology, 65, 355±356. Lau, C.M., Low, L.C., & Eggleton, I.R.C. (1995). The impact of reliance on accounting performance measures on jobrelated tension and managerial performance: additional evidence. Accounting, Organizations and Society, 20, 359±381. Lawrence, P.R., & Lorsch, J.W. (1967). Organization and environment, managing di€erentiation and integration. Boston: Harvard. Lindquist, T.M. (1995). Fairness as an antecedent to participative budgeting: examining the e€ects of distributive justice, procedural justice and referent cognitions on satisfaction and performance. Journal of Management Accounting Research, 7, 122±147. Lindsay, R.M. (1995). Reconsidering the status of tests of signi®cance: an alternative criterion of adequacy. Accounting, Organizations and Society, 20, 35±53. Lindsay, R.M., & Ehrenberg, A.S.C. (1993). The design of replicated studies. The American Statistician, 47, 217±228. Locke, E.A. (1968). Toward a theory of task motivation and incentives. Organizational Behavior and Human Performance, 3, 157±189.

Locke, E.A., & Latham, G.P. (1990). A Theory of Goal Setting and Task Performance. Englewood-Cli€s: Prentice-Hall. Lorange, P., & Scott Morton, M.S. (1974). A framework for management control systems. Sloan Management Review, 16, 41±56. MacDonald, A.P. (1970). Revised scale for ambiguity tolerance: reliability and validity. Psychological Reports, 26, 791±798. Macintosh, N.B., & Daft, R.L. (1987). Management control systems and departmental interdependencies: an empirical study. Accounting, Organizations and Society, 12, 49±61. Macintosh, N.B., & Williams, J.J. (1992). Managerial roles and budgeting behavior. Behavioral Research in Accounting, 4, 23±48. Magner, N., Welker, R.B., & Campbell, T.L. (1995). The interactive e€ect of budgetary participation and budget favorability on attitudes toward budgetary decision makers: a research note. Accounting, Organizations and Society, 20, 611±618. Mak, Y.T. (1989). Contingency ®t, internal consistency and ®nancial performance. Journal of Business Finance and Accounting, 16, 273±300. Martin, J.G., & Westie, F.R. (1959). The tolerant personality. American Sociological Review, 24, 521±528. McConkie, M.L. (1979). A clari®cation of the goal setting and appraisal processes in MBO. Academy of Management Journal, 4, 29±40. Merchant, K.A. (1981). The design of the corporate budgeting system: in¯uences on managerial behavior and performance. The Accounting Review, 56, 813±829. Merchant, K.A. (1982). The control function of management. Sloan Management Review, 23, 43±55. Merchant, K.A. (1984). In¯uences on departmental budgeting: an empirical examination of a contingency model. Accounting, Organizations and Society, 9, 291±307. Merchant, K.A. (1985a). Control in business organizations. Boston: Pitman. Merchant, K.A. (1985). Organizational controls and discretionary program decision making: a ®eld study. Accounting, Organizations and Society, 10, 67±85. Merchant, K.A. (1985). Budgeting and the propensity to create budget slack. Accounting, Organizations and Society, 10, 201± 210. Merchant, K.A. (1987). How and why ®rms disregard the controllability principle. In W.J. Bruns & R.S. Kaplan (Eds.), Field studies in management accounting. Boston: Harvard Business School Press. Merchant, K.A. (1989). Rewarding results: motivating pro®t center managers. Boston: Harvard Business School Press. Merchant, K.A. (1990). The e€ects of ®nancial controls on data manipulation and management myopia. Accounting, Organizations and Society, 15, 297±313. Merchant, K.A., & Riccaboni, A. (1990). Performance-based management incentives at the Fiat group: a ®eld study. Management Accounting Research, 1, 281±303. Merchant, K.A., & Simons, R. (1986). Research and control in complex organizations: an overview. Journal of Accounting Literature, 5, 183±203.

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482 Milani, K. (1975). The relationship of participation in budgetsetting to industrial supervisor performance and attitudes: a ®eld study. The Accounting Review, 50, 274±284. Miles, R.E., & Snow, C.C. (1978). Organizational strategy, structure and process. New York: McGraw-Hill. Murray, D. (1990). The performance e€ects of participative budgeting: an integration of intervening and moderating variables. Behavioral Research in Accounting, 2, 104±123. Neu, D. (1992). The social construction of positive choices. Accounting, Organizations and Society, 17, 223±237. Norton, R.W. (1975). Measurement of ambiguity tolerance. Journal of Personality Assessment, 39, 607±619. Onsi, M. (1973). Factor analysis of behavioral variables a€ecting budgetary slack. The Accounting Review, 122±149. Otley, D.T. (1978). Budget use and managerial performance. Journal of Accounting Research, 16, 122±149. Otley, D.T. (1980). The contingency theory of management accounting: achievement and prognosis. Accounting, Organizations and Society, 5, 413±428. Otley, D.T. (1990). Issues in accountability and control: some observations from a study of colliery accountability in the British coal corporation. Management Accounting Research, 1, 101±123. Otley, D.T. (1994). Management control in contemporary organizations: towards a wider framework. Management Accounting Research, 5, 289±299. Ouchi, W.G. (1977). The relationship between organizational structure and organizational control. Administrative Science Quarterly, 22, 95±113. Ouchi, W.G. (1979). A conceptual framework for the design of organizational control mechanisms. Management Science, 25, 833±848. Ouchi, W.G., & Maguire, M.A. (1975). Organizational control: two functions. Administrative Science Quarterly, 20, 559±569. Parker, L.D., & Lewis, N.R. (1995). Classical management control in contemporary management and accounting: The Persistence of Taylor and Fayol's World. Accounting, Business and Financial History, 5, 211±235. Perrow, C. (1967). A framework for comparative organizational analysis. American Sociological Review, 32, 194±208. Perrow, C. (1970). Organizational analysis: a sociological view. London: Tavistock. Poole, M.S., & Van de Ven, A.H. (1989). Using paradox to build management and organization theories. Academy of Management Review, 14, 562±578. Prendergast, C., & Topel, R. (1993). Discretion and bias in performance evaluation. European Economic Review, 37, 355±365. Puxty, T., & Chua, W.F. (1989). Ideology, rationality and the management control process. In W.F. Chua, T. Lowe & T. Puxty, (Eds.), Critical perspectives in management control (pp. 115±139). Houndmills: MacMillan. Rahman, M., & McGosh, A.M. (1976). The in¯uence of organisational and personal factors on the use of accounting information: an empirical study. Accounting, Organizations and Society, 1, 339±355.

481

Ridgway, V.F. (1956). Dysfunctional consequences of performance measurements. Administrative Science Quarterly, 1, 240±247. Rizzo, J.R., House, R.J., & Lirtzman, S.I. (1970). Role con¯ict and ambiguity in complex Organizations. Administrative Science Quarterly, 15, 151±163. Ronen, J., & Livingstone, J.L. (1975). An expectancy theory approach to the motivational impacts of budgets. The Accounting Review, 50, 671±685. Ross, A. (1994). Trust as a moderator of the e€ect of performance evaluation style on job-related tension: a research note. Accounting, Organizations and Society, 19, 629±635. Ross, A. (1995). Job related tension, budget emphasis and uncertainty: a research note. Management Accounting Research, 6, 1±11. Sathe, V. (1978). The relevance of modern organization theory for managerial accounting. Accounting, Organizations and Society, 3, 89±92. Schmidt, F.L., & Hunter, J.E. (1978). Moderator research and the law of small numbers. Personnel Psychology, 31, 215±232. Schoonhoven, C.B. (1981). Problems with contingency theory: testing assumptions hidden within the language of contingency `theory'. Administrative Science Quarterly, 26, 349±377. Schuler, R.S. (1980). De®nition and conceptualization of stress in organizations. Organizational Behavior and Human Performance, 27, 184±215. Schuler, R.S., Beutell, N.J., & Youngblood, S.A. (1989). E€ective personnel management. St. Paul: West Publishing Company. Selto, F.H., Renner, C.J., & Young, S.M. (1995). Assessing the organizational ®t of a just-in-time manufacturing system: testing selection, interaction and systems models of contingency theory. Accounting, Organizations and Society, 20, 665±684. Shields, J.F., & Shields, M.D. (1998). Antecedents of participative budgeting. Accounting, Organizations and Society, 23, 49±76. Simon, H.A., Kozometsky, G., Guetzkow, H., & Tyndall, G. (1954). Centralization vs Decentralization in organizing the controller's department, New York, NY: The Controllership Foundation. Simons, R. (1987a). Planning, control and uncertainty: a process view. In W.J. Bruns & R.S. Kaplan (Eds.), Accounting and management, ®eld study perspectives (pp. 339±362). Boston: Harvard Business School Press. Simons, R. (1987). Accounting control systems and business strategy: an empirical analysis. Accounting, Organizations and Society, 12, 357±374. Simons, R. (1995) Levers of control. Boston: Harvard Business School Press. Skinner, R.C. (1990). The role of pro®tability in divisional decision making and performance evaluation. Accounting and Business Research, 20, 135±141. Smith, P.C. (1983). Behaviors, results and organizational e€ectiveness: the problem of criteria. In M.D. Dunnette (Ed.), Handbook of industrial and organizational psychology. New York: John Wiley & Sons.

482

F.G.H. Hartmann / Accounting, Organizations and Society 25 (2000) 451±482

Southwood, K.E. (1978). Substantive theory and statistical interaction: ®ve models. American Journal of Sociology, 83, 1154±1203. Swieringa, R.J., & Moncur, R.H. (1972). The relationship between managers' budget-oriented behavior and selected attitude-position, size and performance measures. Journal of Accounting Research, 10, 194±214. Swieringa, R.J., & Moncur, R.H. (1975). Some e€ects of participative budgeting on managerial behavior, New York: National Association of Accountants. Thompson, J.D. (1967). Organizations in action. Social science base of administrative theory. New York: McGraw-Hill. Tosi, H.L. (1975). The human e€ects of managerial budgeting systems. In J.L. Livingstone (Ed.), Managerial accounting: the behavioral foundations. Georgia Institute of Technology. Tosi, H.L., Rizzo J.R., & Carroll S.J. (1994). Managing organizational behavior. Cambridge: Blackwell. Umapathy, S. (1987). Current budgeting practices in U.S. industry. New York: Quorum. Vancil, R.F. (1979). Decentralization: managerial ambiguity by design. Homewood: Dow Jones-Irwin. Venkatraman, N. (1989). The concept of ®t in strategy research: toward verbal and statistical correspondence. Academy of Management Review, 14, 423±444.

Venkatraman, N., & Camillus, J.C. (1984). Exploring the concept of `®t' in strategic management. Academy of Management Review, 9, 513±525. Waterhouse, J.H., & Tiessen, P. (1978). A contingency framework for management accounting research. Accounting, Organizations and Society, 3, 65±76. Williams, J.J., Macintosh, N.B., & Moore, J.C. (1990). Budget-related behavior in public sector organizations: some empirical evidence. Accounting, Organizations and Society, 15, 221±246. Wood, R.E., Bandura, A., & Bailey, T. (1990). Mechanisms governing organizational performance in complex decisionmaking environments. Organizational Behavior and Human Decision Processes, 46, 181±201. Wood, R.E., Mento, A.J., & Locke, E.A. (1987). Task complexity as a moderator of goal e€ects: a meta-analysis. Journal of Applied Psychology, 72, 416±425. Woodward, J. (1965). Industrial organization: theory and practice. London: Oxford University Press. Young, S.M. (1996). Survey research in management accounting: a critical assessment. In A.J. Richardson (Ed.), Research methods in accounting. Vancouver: CGA Research Foundation.