Tài liệu Rockness and shields car 1988

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/229938166 An Empirical Analysis of the Expenditure Budget in Research and Development Article in Contemporary Accounting Research · April 2010 DOI: 10.1111/j.1911-3846.1988.tb00685.x CITATIONS READS 49 69 2 authors: Howard O. Rockness Michael D. Shields University of North Car… Michigan State Univer… 8 PUBLICATIONS 285 69 PUBLICATIONS 3,903 CITATIONS CITATIONS SEE PROFILE SEE PROFILE All content following this page was uploaded by Michael D. Shields on 10 July 2015. The user has requested enhancement of the downloaded file. An empirical analysis of the expenditure budget in research and development"^ HOWARD O. ROCKNESS University of North Carolina at Chapel Hill MICHAEL D. SHIELDS San Diego State University Abstract. This paper extends the prior empirical research that explains the perceived use or importance of budget control or both, with organizational context and structure in manufacturing organizations. In this paper, the perceived importance of expenditure budget control in research and development (R&D) work groups is explained empirically by organizational context (R&D work group size, source of R&D funding, and size of R&D budget) and the management control system (steps in the control process, social control). Data obtained from 76 R&D work groups in ten organizations support the five hypotheses. Generally, there is an interaction between the steps in the control process and each of the other independent variables on the perceived importance of expenditure budget control for management control of the R&D work group. Resume. Le present article s'inscrit dans le prolongement des travaux de recherche empiriques precedents visant k expliquer l'utilisation pergue du controle budgetaire ou son importance - sinon les deux - dans le contexte et la structure d'organisation des entreprises manufacturieres. Dans cet article, l'importance du controle du budget des investissements pergue par les groupes de travail en recherche et developpement (R & D) s'explique concretement par le context organisationnel (la taille dti groupe de travail en R & D, la source de financement des activites de R & D et l'importance des credits affectes a ces activites) et par le systfeme de eontrole de gestion (etapes du processus de controle, controle social). Les donn6es obtenues aupres de 76 groupes de travail en R & D dans dix organisations viennent confirmer les einq hypotheses des auteurs. L'on releve, de fagon generale, une interaction entre, d'une part, les etapes du processus de controle et chacune des autres variables independantes et, d'autre part, l'importance pergue dti controle du budget des investissements aux fins du controle de gestion du groupe de travail en R & D. Introduction Recent empirical research in managerial accounting explains variation in the perceived use and importance of budget control in organizations (Bruns and EDITOR'S NOTE: An earlier draft of this paper and the discussions by Jake Bimberg (see pp. 582-587) and John Waterhouse (see pp. 588-594) were presented at the 1987 CAR Conference on Economic and Behavioral Implications of Accounting Information and its Validation at McMaster University. * The authors would like to thank Jake Bimberg, John Waterhouse and participants at the CAR Conference for their many constructive comments on an earlier draft of this paper. Contemporary Accounting Research Vol. 4 No. 2 pp. 568—581 Expenditure Budget in Research and Development 569 Figure 1 Mapping of organizations by knowledge of transformation process and measurability of output Measurability of output Low High Knowedge of input-output transformation process High R&D Retailing Administration Manufacturing Waterhouse (1975), Khandwalla (1972), Merchant (1981 and 1984)). This prior research primarily examines operating budgets in manufacturing settings. The present paper extends these prior studies by focusing on reseai'ch and development (R&D) and by including new explanatory variables. R&D is an increasingly important component of postindustrial economies. For example, the 844 U.S. companies in the annual Business Week survey on R&D spent an average of 42 percent of pretax profits (or three percent of sales) on R&D in 1985. This was a ten percent increase from 1985 {Business Week (1986)). This research is framed using a contingency model proposed and tested by Ouchi (1977, 1979). Ouchi's framework locates organizations and their subunits on each of two dimensions: (1) knowledge of the process which transforms resource inputs ioto outputs and (2) measurability of the output. The result, in simplified form, is the Iwo-by-two matrix in Figure 1. A manufacturing organization is typical of an organization in which the specific steps necessary to transform resource inputs into outputs are well known and a measure of output is usually available on a timely basis. Flexible budgets are well suited for control of such organizatioos since they are based on fixed rules of performance, such as staadard costs and variance analysis, and departures from standards are observable. Such settings have been the focus of most of the accounting research on budgetary control systems. R&D is an important example of an organization that is at the opposite extreme in Onchi's framework. In R&D, it is very difficult to obtain specific knowledge about flie nature of the transformation process. It is also difficult to obtain measures of output until long after the current activities have taken place. Flexible budgets are not well suited for control in such settings, since they depend on both precise knowledge of the transformation process and timely measurability of outputs. There axe also organizational units that fall between these two extremes. An important example is retail selling organizations. They are characterized by low knowledge of the transformation process, since it is difficult to specify exactly what inputs are necessary, and how to transform those inputs in order to be 570 H.O. Rockness M.D. Shields successful at selling. Nevertheless, outputs (sales) are easily measurable in a timely fashion. A different example is administration, e.g., insurance claims processing. In this case, the transformation process is well known, but the output is difficult to measure since it includes appropriate rejection of unjustified claims as well as customer satisfaction. The present study analyzes differences in the perceived importance of expendittire budget control in R&D work groups that is due to the organizational context (organizational size, expenditure budget size, source of funds) and the management control system (importance of social control, steps in the control process). This organizational setting is in the low knowledge/low measurability ceil of Ouchi's framework (Figure 1). The present research uses Ouchi's framework to provide a classification scheme of organizations and their subunits and implications for control system design. It does not provide a test of variation in the use of controls as a function of the two dimensions of the framework. The first section of the paper reviews prior research on budget control from which the research hypotheses are developed. The second section presents the research method and develops the operational definitions and measurements of the variables utilized in the study. In the third section the results are presented and the final section contains a discussion of the results and directions for future research. Prior research and hypotheses Prior empirical research on budgets explains variation in the perceived importance of budget control based on organizational context and stmcture. Context includes both the organization's extemal environment (e.g., competition, uncertainty) and intemal environment (e.g., size, technology). Structure is the pattem of relationships among people which facilitate accomplishment of an organization's goals, and includes decentralization, formalization and complexity (Gerwin (1981, pp. 3-38); Hall (1982, pp. 49-126)). Bmns and Waterhouse (1975) found that managers believe budget control is more important and is used more frequently in larger, decentralized, more technologically sophisticated organizations in which there are formal and standard operating procedures. Merchant (1981, 1984) found that managers perceive that budget control is both more frequently used and more important in larger, more diverse, more functionally differentiated, decentralized organizations which have automated technology. Khandwalla (1972) reports that managers' perceived use of flexible budget control is a positive function of competition that confronts their organization. Prior research also examined the perceived use or importance of operating or flexible budgets. In this research, the expenditure budget is analyzed, since prior research indicates that it is the type of budget used in R&D (Anthony (1952, pp. 138-140); Anthony, Dearden and Bedford (1984, p. 207); Gambino and Gartenberg (1979, pp. 7, 9); Lin and Vaserhelyi (1980)). In addition, the expenditure budget is probably frequently used by other organizations or organizational units in which there is a low level of knowledge of the input-output transformation process and where outputs are difficult to measure on a timely Expenditure Budget in Research and Development 571 basis. Examples include administration, legal and advertising. Expenditure budget control is characterized by setting limits on the dollar size of activities, and it serves to authorize the amount of funds that an organizational unit can spend to support its activities. As such, expenditure budget control is primarily exercised during planning to set the scope of activities and to authorize spending on those activities. It also is used to monitor whether spending is within authorized guidelines. Management control system Control process The importance of expenditure budget control has been argued to vary with the particulair step in the management control process (Anthony (1952, pp. 162-163, 183-184); Anthony, Dearden and Bedford (1984, p. 201); Gambino and Gartenberg (1979, pp. 7, 9, 120)). Anthony, Dearden, and Bedford (1984, pp. 10-28) suggest that the four chronological steps in the control process for R&D are planning, monitoring, evaluating, and rewarding. For example, an expenditure budget is argued to be important as a control for planning because it details the flow of resources which then guides management's action (Anthony, Dearden and Bedford (1984, p. 201)). The expenditure budget is also important for monitoring resource flows since it indicates how close those flows are to the authorized spending level. The importance of expenditure budget control is likely to decrease for steps in the control process that are more removed fi-om planning, such as evaluatiHg and rewarding. One reason is that the expenditure budget, unlike a flexible budget, generally is not ex post tailored to the actual level of activity. Consequently, the expenditure budget is unlikely to be as important as a flexible budget for evaluating and rewarding, since, like a static budget, it provides a rather crude standard against which to compare actual spending (Anthony, Dearden and Bedford (1984, p. 201)), There is no published literature which empirically tests whether the perceived importance or use of (expenditure) budget control varies with the steps in the control process. HI: The perceived importance of expenditure budget control in R&D varies with the chronological steps in the management coEtrol process. Specifically, the level of perceived importance decreases motiotonically from planning to monitoring, monitoring to evaluating, and evaluating to rewarding. Social control Prior empirical research on budget control in organizations indicates that budget control and social control are used with different organizational contexts and structures (Bruns and Waterhouse (1975); Merchant (1981)). Social control can be characterized by a rigorous personnel selection policy, hazing, direct supervision, oral commuDication, continual peer interaction and professional association. The importance of social control in R&D management has been argued for and empirically demonstrated by prior research (Anthony (1952, pp. 193, 333-334); 572 H.O. Rockness M.D. Shields Gambino and Gartenberg (1979, pp. 106-107); Waterhouse and Tiessen (1978)). Social control tends to be more prevalent in smaller, centralized, technologically simple organizations (Bnins and Waterhouse (1975); Merchant (1981, 1984)). In contrast, as reviewed earlier, budget control often is more prevalent in larger, decentralized, technologically sophisticated organizations. Thus, these controls are effective within different organizational contexts and structures. Although both may be present, emphasis may be placed more strongly on one than the other. This is hypothesized to hold despite the emphasis in R&D environments on professional behavior. Thus, social control and expenditure budget control may be substitutable because each is likely to be effective aod efficient in different situations. H2: There is a negative association between the perceived importance of social control and the perceived importance of expenditure budget control. Organizational context Prior research has established that organizational context variables are associated with variation in the perceived use or importance of budget controls (Bruns and Waterhouse (1975); Khandwalla (1972); Merchant (1981, 1984)). Although none of this research has focused on R&D, it has established the linkage between organizational context and budget use at the organizational level and, by implication, at unit levels within an organization. Hayes (1977) found contingency relationships in performance measurement aod evaluation among organizational subunits, specifically R&D, production and marketing. In this paper, the hypotheses are developed based on the existing literature; however, the level of analysis is explicitly identified as the R&D work group. The organizational context variables included in this research are size of the work group, source of R&D funds, budget size, measurability of the R&D output, aod knowledge of the R&D technology. The latter two variables are included to extract statistically any residual variance in the data that is due to variation in the measurability of outputs and knowledge of transformation process (see Ouchi's framework in Eigure 1). Thus, no hypotheses are offered for these two variables as they serve the role of covariates in the hypothesis testing. Work group size The size of the work group is included since the organizational behavior literature indicates that organizational size explains significant variation in formal organizational structures such as (expenditure) budget control (Eord and Slocum (1977); Hall (1982, p. 56); Kimberly (1976)). As an organization gets larger, it typically demands more formal stmctures to deal effectively with the increased complexity of communication aod coordination (Bruos and Waterhouse (1975); Merchaot (1981, 1984)). It is expected that these prior results apply to R&D work groups. H3: There is a positive association between the perceived importance of expenditure budget control and the R&D work group size. Expenditure Budget in Research and Development 573 Source of R&D funds Zimmerman (1976) provides empirical evidence that the level of uncertainty conceming the source of funding, internal or external, for R&D affects management's behavior with respect to the pattern of spending on R&D. Since external sources are often more uncertain both as to timing of receipt and amount, related e:ii;penditures tend to be discontinuous and thus are less likely to be efficient and effective. Further, as Zimmerman argues, management has more incentive to spend within the constraints of a budget when the source of funds is external, relative to internal, because of the greater cost of having a negative relationship with a funding source. This is likely to make the expenditure budget more important for control when the work group receives external funding. The expeaditiire budget also may be more important for control as more funds are obtained externally, since the funding source often requires that it be used as a meaes of monitoring the use of external funds. H4: There is a positive association between the perceived importance of expendittire budget control and the relative amount of funds received by the R&D work group from sources external to the organization. Size of budget Anthony, Dearden and Bedford (1984, p. 200) argue that a principal way in which maBagement controls a discretionary expense center such as R&D is by planning the size or scope of jobs, and a common way to represent size or scope is the dollar size of the (expenditure) budget. Consequently, the perceived importance to managenaeot of a job or organizational unit, such as the R&D work group, may be proxied by the dollar size of the expenditure budget. Similarly, an R&D manager may believe that the expenditure budget is more important as the dollar size of the badget increases because higher-level management allocates its attention to jobs or units according to their perceived importance to the organization. More directly, the expenditure budget may be more important as its size increases because of the potential effect of the organizational unit on the firm's performance and maoagement's desire for a common denominator, such as a budget, to control its larger activities. H5: There is a positive association between the perceived importance of expenditure budget control and the dollar size of the R&D work group budget. Hesearch method Sample The unit of analysis in this research was an organizational work group headed by an R&D supervisor. Questionnaires were distributed by R&D executives in each participating organization to a total of 113first-lineR&D supervisors. Seventy-six (67 percent) usable questionnaires were returned by mail. Respondents returned 574 H.O. Rockness M.D. Shields their questionnaires directly rather than through the executive distributing the questionnaires in the organization. As a measure of the respondents' interest, fifty (66 percent) of them asked for a copy of the results. The sample was drawn from ten organizations, of which eight organizations were private, for-profit corporations, one was a federal govemment organization and the other was a private, not-for-profit corporation. Five of the organizations were located in a large R&D park in North Carolina, while the others were dispersed throughout the U. S. These organizations conducted R&D in thefieldsof electronics, communications, computers, environment, forest products, pharmaceuticals, toxicology, chemistry, containers, and research in the physical and social sciences. A five-page questionnaire was answered by each respondent. The first page provided the context of the research and definitions of key terms. The rest of the questionnaire was divided into four sections. The first requested information on the respondent's work group and organization. The second section requested information on the work group's budgeting and funding. The third set of questions asked for information on work group tasks, while the fourth section elicited information on work group controls. It was not. possible to gather information for analysis of differences between organizations, since the ten organizations required full anonymity as a condition of participation. The R&D organizations in the sample were very sensitive on the issue of confidentiality. The result was that there was no ethical way to systematically assess the presence of a nonresponse bias. However, responses were received promptly after distribution with no apparent pattem of differences in responses. In addition, the response rate was very high (67 percent) for this type of research and those requesting copies of results (66 percent) were distributed across the ten organizations. Thus the potential for nonresponse bias is believed to be small, although certainly a potential limitation on the generalizability of the results. The requirement for anonymous responses also prevented a second request to nonrespondents for completion of the questionnaire. Measurement of variables Management control system Prior empirical research on budget and social control has operationalized and measured controls as perceived use and/or importance for managing an organization (Brans and Waterhouse (1975); Khandwalla (1972); Merchant (1981,1984)). In some studies, one scale was used to measure the overall use or importance of a control in managing. In others, overall use or importance was measured on subscales for one or more of the various steps in the management control process, such as planning, monitoring, evaluating and rewarding, and then these subscales were aggregated. In this research, the perceived importance of each control was measured for each step in the management control process. The primary difference in measurement between the present and prior research was the response format. In Expenditure Budget in Research and Development 575 order to- provide a respondent with the opportunity to have a direct and systematic response to the perceived importance of each control for each step of the management control process, each response scale was in a cell of a twodimensional matrix. The column dimension included expenditure budget control and social control. The row dimension was comprised of the four steps in the management control process: planning, monitoring, evaluating and rewarding. The importance ratings were on a five-point scale: no (0), little (1), moderate (2), considerable (3), and extreme (4). Organizational context The size of the work group was measured as the sum of professional or technical persons in the work group in each of the following fields: arts/humanities, biological sciences, engineering, medical, quantitative methods/computer science, physical sciences, social sciences and other. While the "other" category was included to ensure exhaustive measurement, it did not account for a significant number of personnel. The size of the expenditure budget was measured as the total number of dollars received by the work group. Five intervals were used including: under $150,000; $150,000 to $300,000; $301,000 to $600,000; $601,000 to $1, 200,000; and over $l,200,i000. Intervals were used both to encourage responses from respondents whose organizations might be reluctant to disclose the exact size of their budget aod for ease of response. Source of R&D funds was measured by having respondents enter the percentage of funds received from within their organization, from governmental organizations and from other extemal sources. The sum of the percentages entered in the latter two categories was used as the measure of extemal source of funds. The timelifless of the measurability of the output was captured by the questions: "How soon is your work group's output measurable in numbers?" (anchored by (1) "Immediately measurable" and (9) "Never measurable") and "To what extent is your work group's output measurable in numbers?" (anchored by (1) "Not measurable in numbers" and (9) "Completely measurable in numbers"). Knowledge of the R&D input-output transformation process was measured as: "How much do you know about the research and/or development transformation process your work group uses to yield a specific output?" (anchored by (1) "No knowledge" and (9) "Complete knowledge") and as "To what extent do you know that the set of inputs your work group uses will result in a specific output?" (anchored by (1) "No knowledge of relationship between input and output" and (9) "Complete knowledge of relationship between input and output"). ResHlts Descriptive statistics Means and standard deviations of the dependent and independent variables are preseoted in Tables 1 and 2. Inspection of the data indicated that the budget size 576 H.O. Rockness M.D. Shields TABLE 1 Descriptive statistics Variable Mean Std, dev. Budget - Planning Budget - Monitoring Budget - Evaluating Budget - Rewarding Social - Planning Social - Monitoring Social - Evaluating Social ~ Rewarding Work group size External source funds* Size of R&D budget* 2.96 2.66 2,28 2,11 2,25 2,29 2,26 2.36 18.95 0,20 0,50 1,07 1,02 1.04 1,15 1,21 1,24 1.25 1.21 23,70 0.40 0.50 74 74 74 73 73 73 73 74 73 76 75 * 0-1 dummy variable. TABLE 2 Perceived importance of expenditure budget control Control process Source of funds Size of R&D budget Social control Work group size Low High Small Large Int, Ext. Small Large Planning X 3,24 0,76 2,67 1.29 2,91 1,13 2,97 1,06 2,91 1,16 3,20 0,68 2,89 1,16 3,06 1,01 Monitoring X 2,73 0,84 2,61 1,25 2,42 1.06 2,79 0,98 2,56 1,03 3,07 1,03 2,49 1,01 2,82 1,06 Evaluating X 2.49 0,93 2,12 1,14 1.91 1,07 2.59 0,86 2,22 1,03 2,67 1.05 1,83 0,98 2,76 0,85 1.92 1.09 2.24 1,20 1,76 1,00 2,32 1,25 1,98 1.03 2,40 1,50 2,06 1,11 2,06 1,20 s Rewarding X s Legend: x = mean; s = standard deviation. and external source of R&D funds variables were significantly skewed. Consequently, they were rescaled to be 0-1 dummy variables. Zero represented either small budget size or internal source(s) of funds, while one represented either large budget size or external source(s) of funds. Budget size was split at the median, with the result that small budget size was defined as a budget of less than or equal to $1,200,000. The source of funds variable was bimodal, since work groups received either almost all (> 97 percent) or none (^ 2 percent) of their funds from internal sources. Internal sourcing of funds was defined as greater than or equal to 97 percent of funds being received from internal sources and external sourcing of funds was defined as greater than or equal to 98 percent of funds being received from external sources. This classification scheme was intended to separate categorically those work groups which had significant external funds from those which did not. Expenditure Budget io Research and Development 577 Hypothesis testing HI was tested by a four-level one-way repeated measure analysis of covariaoce (ANCOVA). The dependent variable was budget importance, the four levels were the steps in the control process while the covariates were the four measurability of output and knowledge of R&D transformation process variables. The other four hypotheses were each tested by a two by four ANCOVA. The dependent variable, again, was budget importance. The four levels of the repeated measures variable were the steps in the control process with the two levels of the between-subjects variable dependent on the specific hypothesis (low or high social control, small or large work group size, internal or external source of funds, small or large budget size). In testing H2, social control was dichotomized by splitting it at its median; the result was that low social control was less than nine and high social control was greater than nine. In testing H3, work group size was dichotomized by splitting it at its raedian; the result was that small size was less than 11 professional employees and large size was greater than or equal to 11 professional employees. The covariates were the four measurability of output and knowledge of transformation process variables. None of the covariates in aoy of the five ANCOVAs was statistically significant at conventional levels of significance. Management control system Control process Ul predicted that the perceived importance of the expenditure budget for management controi io R&D moootonically decreases across planoiog, monitoring, evjiluating and rewarding. The hypothesis was tested using a one-way repeated measures ANCOVA with Tukey contrast analysis. The perceived importajice of expenditure budget control was significaotly differeot as expected across steps io the control process (F = 14.1, p < 0.0001). (This cootrol process variable was also highly significant (p < 0.001) in each of the ANCOVAs to be reported for the tests of H2-H5.) The means were 2.96 for planning, 2.66 for monitoring, 2.28 for evaluating and 2.11 for rewarding (see Table 1). The contrast analysis indicated that five of the six pairwise contrasts were sigoificaot (p < 0.05), the exception was the difference between evaluating aod rewarding. These results supported HI. Support for E l raised the potential for interaction effects between steps io the control process and each of the other blocking variables. For example, H2 suggested that there would be a negative association between the perceived importance of social control and the perceived importance of budget control. Results for HI suggested that the degree of association between the importance of budget control and social control may be contingent on the particular step in the contj-ol process. Statistically, this outcome would be observed if there was a significant interactioo effect between steps in the control process and the importance of social cootrol. Thus, interaction effects are examined in the results which follow. 578 H.O. Rockness M.D. Shields Social control The mean of social control was invariant to the step in the control process (see Table 1): planning, 2.25; monitoring, 2.29; evaluating, 2.26; rewarding, 2.36. H2 predicted a negative association between the perceived importance of expenditure budget control and social control. Thus, we expected to find that budget control was more (less) important when social control was of low (high) importance. Inspection of Table 2 found this result for three of the four control steps; the exception was rewarding. Across the steps in the control process, the differences in means were 0.57 for planning, 0.12 for monitoring, 0.37 for evaluating, and -0.32 for rewarding. The importance of expenditure budget control varied between levels of social control almost five times (0.57/0.12) as much for planning as for monitoring. This indicated that, for monitoring, the expenditure budget had a fairly constant level of importance regardless of the importance of social control. Apparently, work groups perceived the expenditure budget always to be important for monitoring. For rewarding, the expenditure budget was more important when social control had high vs. low importance (2.24 vs. 1.92). Consequently, since the difference in the importance of budget control varied between levels of social control as well as among steps in the control process, there was a significant interaction (F = 3.36, p = 0.02) between steps in the control process and social control, but not a significant main effect for social control (F = 1.19, p = 0.28). The nonsignificant main effect for social control was due to the expenditure budget being less important for rewarding when the level of importance for social control was high vs. low. The effect was to reduce the marginal difference in means between levels of social control. In this case, the interaction effect provides the test of the hypothesis. Thus, these results supported H2. Organizational context Work group size H3 predicted a positive association between expenditure budget control and work group size. For all four steps in the control process, the importance of budget control was higher when the work group was larger (see Table 2): 0.06 higher for planning; 0.37 for monitoring; 0.68 for evaluating; and 0.56 for rewarding. Except for planning, budget control was more important as work group size increased. The budget was uniformly of relatively high importance for planning. The inferential result was a significant (F = 6.07, p < 0.02) main effect for work group size and a marginally significant (F = 2.22, p = 0.095) interaction between steps in the control process and work groups size. Overall, these results supported H3. Source of R&D funds H4 predicted that budget control is more important when funds are externally sourced. For all four steps in the control process, the expenditure budget was more important when funds were from external compared with internal sources (see Expenditure Budget in Research and Development 579 Table 2). The differences in means due to sourcing for the four steps were fairly uniform ranging from 0.29 for planning to 0.51 for monitoring. There was a significant main effect due to source of funds {F = 3.50, p = 0.06) but not a significant interaction between source of funds and steps in the control process {F = 0.17, p = 0.92). Overall, these results provided support for H4. Size of R&D budget H5 predicted that the importance of budget control increases as the size of the budget increases. For each of the control steps, the budget was more important when its size was larger; the increase in importance was 0.17 forplanning; 0.33 for monitoring: 0.93 for evaluating; and 0.002 for rewarding. Since the difference in means varied with the steps in the control process, there was a significant interaction between control process and budget size {F = 5.59, p = 0.002). The budget size main effect was also significant (^F = 4.41, p = 0.04). These results supported H5. ' ,., Discussion and conclusions The results and following conclusions should be interpreted within the caveats associated with exploratory, cross-sectional field survey research where data have been collected from a limited number of organizations. A primary result of the tests of liie hypotheses is that, in general, the importance of expenditure budget control was explained statistically by the management control system - steps in control process (HI) and the importance of social control (H2) - and organizational context variables - work group size (H3), source of R&D funds (H4), and budget size (H5). In addition, there were interactions between the steps in the coBtroJ process and each of social control, work group size and budget size, but not with source of funding. The results of this study both support and extend the prior research on budget control. These results support prior research by demonstrating a contingency relationship between budget control and organizational context. The importance of expenditure budget control was found to vary with three context variables, work group size, budget size and source of R&D funds. The results of this study also extend the prior research in three ways. First, a new budget a ad a new functional area were examined. This extension is valuable given the increasing importance of R&D in society. As society shifts away from organizations in which flexible budgets are appropriate toward organizations in the low/low cell of Ouchi's fraroework (see Figure 1), additional research on organizational units, such as administration, services, retailing and R&D, is valuable in understanding budget control in contemporary organizations. Second, the results show an interaction between steps in the control process and each of social control, work group size and budget size. Future research should explore the extent to which variation in the importance of the budget across steps in the control process is unique to the expenditure budget. The control process may have beeo important because the expenditure budget is not ex post tailored to 580 H.O. Rockness M.D. Shields actual activity and thus it is more important for early steps in the process. In contrast, a flexible budget may be of universal importance, or even of increased importance, through the steps in the control process, because it is ex post tailored to a specific level of activity. Third, our results as well as those of Bruns and Waterhouse (1975) and Merchant (1981) indicate that budget control and social control are used in different organizational settings and may be partially substitutable. Future research should examine further the contingent nature of both the mix and use of altemative controls. Future research could examine both additional variables and address the measurement issues inherent in both this and the prior budget research. The additional variables might include strategy, incentives, incentive risk sharing trade-offs, short run vs. long run balance, peformance, control by reputation and culture, and intemal competition for funds. The breadth and exploratory nature of this research precluded formal attempts to address measurement issues here. This research relied on measurements which had been used in prior research. The generaUzability of the results is limited to the extent that there is measurement error. Future research should address more carefully the issues of validity and reliability of the measurement of the variables. Such future extensions are a necessary part of developing a more complete understanding of budget control. References Anthony, R., Management Controls in Industrial Research Organizations (Harvard University Press: Boston, Mass. 1952). - , J. Dearden and N. Bedford, Management Control Systems (Irwin: Homewood, Illinois 1984). Bruns, W. and J. Waterhouse, "Budgetary Control and Organizational Structure," Journal of Accounting Research (Autumn 1975) pp. 177-203. Business Week, "R&D Scoreboard" (June 23, 1986) pp. 134-56. Ford, J. and J. Slocum, "Size, Technology, Environment and the Stmcture of Organizations," Academy of Management Review (October 1977) pp. 561-75. Gambino, A. and M. Gartenberg, Industrial R&D Management (National Association of Accountants: New York, New York 1979). Gerwin, D., "Relationships Between Structure and Technology," in: P. Nystrom and W. Starbuck (eds.). Handbook of Organization Design (Oxford University Press: Oxford, England 1981) pp. 3-38. Hall, R., Organizations: Structure and Process, Third Edition (Prentice-Hall: Englewood Cliffs, New Jersey 1982). Hayes, D., "The Contingency Theory of Managerial Accounting," The Accounting Review (January 1977) pp. 22-39. Khandwalla, P., "The Effects of Different Types of Competition on the Use of Management Controls," Journal of Accounting Research (Autumn 1972) pp. 275-85. Kimberly, J., "Organizational Size and the Structuralist Perspective: A Review, Critique and Proposal," Administrative Science Quarterly (December 1976) pp. 571-97. Lin, T. and M. A. Vasarhelyi, "Accounting and Financial Control for R&D Expenditures," TIMS Studies in the Management Sciences (Number 15 1980) pp. 199-213. Merchant, K., "The Design of the Corporate Budgeting System: Influences on Managerial Behavior and Performance," The Accounting Review (October 1981) pp. 813-29. Expenditure Budget in Research and Development 581 , "Influences on Departmental Budgeting: An Empirical Examination of a Contingency Model," Accounting, Organizations and Society (June 1984) pp, 291-307. Ouchi, W., "The Relationship Between Organizational Structure and Organizatiooal Control," Adminstrative Science Quarterly (March 1977) pp. 95-113. , "A Conceptual Framework for the Design of Organizational Control Mechanisms," Management Science (September 1979) pp. 833-47, Waterhouse, J. and P. Tiessen, "A Contingency Framework for Management Accounting Systems Research," Accounting, Organizations and Society (August 1978) pp. 65-76. Zimmennan, G., "Budget Uncertainty and the Allocation Decision in a Nonprofit Organization," Journal of Accounting Research (Autumo 1976) pp. 301-19,