Strategic Decision-Making Processes The Role of Management and Context_甜梦文库
Strategic Decision-Making Processes The Role of Management and Context
Strategic Decision-Making Processes: The Role of Management and Context Author(s): Vassilis M. Papadakis, Spyros Lioukas, David Chambers Source: Strategic Management Journal, Vol. 19, No. 2 (Feb., 1998), pp. 115-147 Published by: John Wiley & Sons Stable URL: http://www.jstor.org/stable/3094060 Accessed: 01/09/Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=jwiley. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Strategic Management Journal.http://www.jstor.orgStrategic Management Journal, Vol. 19, 115-147 (1998)KSTRATEGIC DECISION-MAKING PROCESSES:THE ROLEOF MANAGEMENT AND CONTEXTVASSILIS M. PAPADAKIS1*, SPYROS LIOUKAS2 and DAVID CHAMBERS1 'London Business School, London, U.K. 2Athens University of Economics and Business, Athens, GreeceThis paper investigates the relationship between the process of strategic decision-making and management and contextual factors. First, drawing on a sample of strategic decisions, it analyzes the process through which they are taken, into seven dimensions: comprehensiveness/rationality, financial reporting, rule formalization, hierarchical decentralization, lateral communication,politicization, problem-solving dissension. Second, these process dimensions are related to (I) decision-specific characteristics, both perceived characteristics and objective typologies of strategic decisions, (2) top management characteristics, and (3) contextualfactors referring to external corporate environmentand internalfirm characteristics. Overall, the results support the view that strategic decision processes are shaped by a multiplicity of factors, in all these categories. But the most striking finding is that decisionspecific characteristics appear to have the most importantinfluence on the strategic decisionmakingprocess, as decisions with differentdecision-specific characteristics are handled through differentprocesses. The evident dominance of decision-specific characteristics over management and contextualfactors enriches the traditional 'external control' vs. 'strategic choice' debate in the area of strategic management. An interpretation of results is attempted and policy implications are derived. ? 1998 John Wiley & Sons, Ltd. Strat.Mgmt.J., Vol. 19, 115-147 (1998)INTRODUCTIONStrategic decision-making has emerged as one of the most active areas of current management research. The area has greatly benefited from such research traditions as behavioral decision theory and transaction cost economics and has recently gained its own momentum (Schwenk, 1995). However, despite a substantial body of literature, it is still widely recognized that our knowledge of strategic decision-making processes is limited and is mostly based on normative or descriptive studies and on assumptions most of which remain untested (e.g., Bateman and ZeitKey words: strategic decision- top management * Correspondence Vassilis M. Papadakis, to: LondonBusiness School,Sussex Place,Regent'sPark,LondonNW1 4SA, U.K. CCC / $17.50haml, 1989; Langley, 1990; Pettigrew, 1990; Rajagopalan, Rasheed, and Datta, 1993; Rajagopalan et al., 1997; Schneider and De Meyer, 1991). As Eisenhardt and Zbaracki put it, despite the crucial role of strategic decisions, the strategy process research has not departed significantly from a stage of being based on 'mature paradigms and incomplete assumptions' (Eisenhardt and Zbaracki, 1992: 17). In particular, the need has been recognized for integrative research which explicitly considers the impact of context on strategic processes (Bateman and Zeithaml, 1989; Bryson and Bromiley, 1993; Rajagopalan et al., ; Schneider and De Meyer, 1991; Schwenk, 1995). For instance, Pettigrew (1990) asked whether the nature of the decision problem shapes the process more than does the organizational context through which the process proceeds. In the same vein, Rajagopalan Received1 December1994 Revised25 July 1995, 24 December1996 Final revisionreceived5 March 1997? 1998 JohnWiley & Sons, Ltd.116V. M. Papadakis, S. Lioukas and D. Chamberset al. (1997) suggest as one of the priorities of future research in strategic decision-making the examination of the extent to which variations in strategic decision-making processes (DMPs) are explained by variations in organizational, environmental, decision-specific, and managerial factors. The wider literature on environmental determinism and the role of management choice is relevant here. Strategic decisions (SDs) are among the main means through which management choice is actually effected. But empirical research has not been extended to rigorous investigation of the role of management factors, contextual factors, and decision-specific characteristics on the actual strategic decision-making processes. The present paper attempts to contribute to this area drawing upon an in-depth empirical investigation of a number of strategic decisions. Specifically, it focuses on SDs of an investment nature. These are decisions leading to significant commitment of resources, with significant impact on the firm as a whole and on its long-term performance (Marsh et al., 1988). First, the paper analyzes the process through which organizations arrive at an SD. Using prior research and empirical evidence it identifies and measures significant generic dimensions of the process. The dimensions extracted refer to comprehensiveness/rationality, formalization, configuration of the process, and politicization. Second, these dimensions are related to a number of factors belonging to the following categories: decision-specific characteristics, top management characteristics, contextual factors, i.e., external corporate environment, and internal firm characteristics (such as systems, performance, size, ownership). The structure of the paper is as follows. First, we review the theoretical background and propose an integrated research framework for studying the effect of management and context on SD processes. Then comes our research methodology, the consideration of the dimensions of the process of arriving at SDs and the explanation of the selection and operationalization of management and other contextual variables. Next, we present the analysis of the data as well as the main results of the study. Finally, we discuss our results, summarize the main conclusions and derive theory and policy implications.THEORETICAL FRAMEWORKDimensions of SD processes Various dimensions/aspects of SD-making processes have been emphasized in the literature. Many studies in the field of SD-making describe the process as a sequence of steps, phases or routes (e.g., Fredrickson, 1984; Mintzberg, Raisinghani, and Theoret, 1976). Others focus on process dimensions instead (e.g., Bourgeois and Eisenhardt, 1988; Hickson et al., 1986; Lyles, 1987; Miller, 1987; Stein, 1980). Several dimensions of SD processes can be derived from the literature. These include the following: * Comprehensiveness/rationality dimension (Dean and Sharfman, 1993a, 1993b; Lyles and Mitroff, 1980; Miller, 1987). Elements of rationality can also be traced in studies addressing such dimensions as complexity of methodology (Langley, 1990), degree of inquiry (Lyles, 1987), and scrutiny (Cray et al., 1988). * Centralization (Cray et al., 1988; Lyles, 1987; Miller, 1987). * Formalization/standardization of the process (e.g., Stein, 1980). * Political/problem-solving dissension dimension. This includes among others politicality (Lyles, 1987; Hickson et al., 1986; Dean and Sharfman, 1993b; Pfeffer and Salancik, 1974), and negotiation/bargaining (Cray et al., 1988; Hickson et al., 1986; Pettigrew, 1973). * Other factors have also been suggested such as dynamic factors (Cray et al., 1988; Mintzberg et al., 1976), forcing (Bryson and Bromiley, 1993), and duration (Hickson et al., 1986; Wally and Baum, 1994). Characterizationof the DMP on these dimensions allows the researcher to examine possible interrelationships with contextual and other factors. The role of broader context in strategic decision-making Many researchers have referred to aspects of contextual influence on strategic DMPs (e.g., Beach and Mitchell, 1978; Billings, Milbum, and Schaalman, 1980; Bryson and Bromiley, 1993; Dutton, Fahey, and Narayanan, 1983; Hitt and Tyler, 1991; Rajagopalan et al., 1993). Schneider and De Meyer (1991), in an attempt to provide? 1998 John Wiley & Sons, Ltd.Strat. Mgmt. J., Vol. 19, 115-147 (1998)Processes StrategicDecision-making117an integrative model, proposed the following issue identificationand diagnosis (e.g., Billings categorizationof factors which are expected to et al., 1980; Dutton, 1986; Jackson and Dutton, influencestrategicprocesses:(1) managers'indi- 1988). The authorsare not awareof any empirical vidual characteristic (2) work that empirically examines a range of internal or and (3) environ- decision-specificcharacteristics relationshipto in mentalfactors. Pettigrew(1990) suggests that in a range of process dimensions.With few excepadditionto context, researchshould consider the tions (e.g., Dean and Sharfman,1993a; Dutton, role and significance of the nature of the decision 1986; Dutton, Walton, and Abrahamson,1989; Dutton et al., 1983; Fredrickson,1985), existing problem in shapingthe process. An integration these contextualdomainsinto research has not yet shown in any detail how of a wider framework looks a promisingavenue for decision-specificcharacteristicsshape the DMP research. Such a framework mustcombineat least as a whole. the following basic perspectives:an 'individual decision perspective', 'strategic or management The strategic or management choice perspective and choice', 'environmental determinism', a 'firm characteristics and resource availabilityperspec- This perspectiveemphasizesthe role of decisiontive'. The following paragraphsbriefly discuss makers. It stresses that strategic choices have the theoreticalunderpinnings each perspective, an endogenousbehavioralcomponent,and partly of as well as the most importantrelevant research reflect the idiosyncrasies of decision-makers efforts undereach perspective. (Child, 1972; Cyertand March,1963). A number of studies extend this argument further, contending that the role of 'upperechelons' or 'top The decision perspective managers'or 'strategic leadership' is important The nature of the decision itself, or the SD enough to determinestrategycontentand process and project, may be important. Research into (Child, 1972; Hambrick Mason, 1984; Miller decision-makingcognition and labeling suggests and Toulouse, 1986). Researchhas mainly focused on the influence that the same internalor external stimulus may be interpreted quite differently by managersin of top management(i.e., CEO and/or top mandifferent organizationsor even within the same agementteam) on corporate strategies(Miller and organization(e.g., Dean and Sharfman, 1993a; Toulouse,1986;Finkelsteinand Hambrick,1990), Dutton, 1993; Haley and Stumph, 1989). It has on performance (Haleblianand Finkelstein,1993; been argued that the way managerscategorize Smith et al., 1994) and on planning formality and label a decision in the early stages of the (Bantel, 1993). There has been little empirical DMP strongly influencesthe organization'ssub- work on the link between top managementand sequent responses (Dutton, 1993; Fredrickson, the process of making SDs (Bantel, 1993; Huff et 1985; Mintzberg al., 1976). For example,there and Reger, 1987; Lewin and Stephens, 1994; is evidence that if a decision is perceived as a Smith et al., 1994). As Rajagopalan et al. crisis differentactions will be taken than if the () stress in a recent review: decision is perceivedas an opportunity (Jackson factors such as researchrelating organizational and Dutton,1988;Milbur, Schuler,and Watman, team (TMT) characteristics ... top management 1983). Fredrickson (1985) found that when to strategicdecision processesis limited. as decisions were interpreted threatsas opposed the to opportunities, DMP followed was characMoreover, the few studies which have been done on the links between top management terized by greatercomprehensiveness. and Our understanding, however, of the impact of characteristics strategicDMPs have produced on characteristics organizational mixed results. Recently Hitt and Tyler (1991) decision-specific decision-makingprocesses is still quite limited found that the demographic characteristicsof (Papadakis and Lioukas, 1996; Rajagopalanet CEOs (i.e., type of academic education) influal., 1993). Most of the empiricalwork focuses enced the modes of strategic decision-making on: (1) single decision-specific characteristics followed. It is interestingto note that counteror (e.g., opportunity crisis) and theirinfluenceon argumentshave also been advanced. Stein, in aspects of the DMP; or (2) the early stages of studying the strategic DMP, went so far as to? 1998 John Wiley & Sons, Ltd. Strat. Mgmt. J., Vol. 19, 115-147 (1998)118V. M. Papadakis, S. Lioukas and D. Chambersments effective firms follow more rational DMPs. Sharfman and Dean (1991) argued for a link between environmental heterogeneity and standardization in the making of SDs. In a similar vein, Priem, Rasheed, and Kotulic (1995) have found that comprehensive processes led to better performance in rapidly changing environments. In sum, the results of this body do not help us in making any meaningful generalizations (Sharfman and Dean, 1991). Rajagopalan et al. () and Dess and Rasheed (1991) note that the small number of studies adds to the uncertainty as to the effects of each environmental aspect on the process of making SDs. Another criticism is that most research seems to focus mainly on one importantenvironmental characteristic (i.e., environmental uncertainty). Other important characteristics such as environmental munificence-hostility seem to have received The environmental determinism perspective somewhat less attention (Rajagopalan et al., to environmental determinism, stra- 1993). According tegic decisions and processes are expressing adaptation to opportunities, threats, constraints, and The firm characteristics and resource other characteristics of the environment. The role availability perspective of top managers is minimized to a facilitation of this adaptation. Hannan and Freeman (1977) and This perspective emphasizes internal factors such Aldrich (1979) go even further to propose a as: internal systems, company performance, size, process of natural selection of species for organi- corporate control (i.e., ownership). At the level zations: the environment determines who will sur- of theory, it can be linked to the 'inertial' pervive, while top managers are passive agents with spective proposed by Romanelli and Tushman minimal impact on corporate development. This (1986), according to which existing organiview is in line with economic theories in which zational arrangements, structures, systems, procdecision outputs rather than internal DMPs are esses, and resources, though initially determined relevant for the explanation of a firm's behavior by management and environmental forces, in turn in a competitive environment. constrain future strategic decision-making. It is In the context of SDs the environmental deter- also related to resource availability such as profminism perspective mainly addresses the question itability and slack resources. More specificially: of how environmental factors (e.g., dynamism, hostility) influence strategic DMPs. Few empirical Internal systems. The systems of an organization studies can be found here (e.g., Fredrickson, (especially formal planning systems (FPSs), 1984; Eisenhardt, 1989; Judge and Miller, 1991) might be expected not only to exert significant and those available seem to have produced con- influence on the flow of information between the tradictory results (Rajagopalan et al., 1997). For layers of hierarchy, but also to determine the example, Fredrickson and Iaquinto (1989) con- nature and context of human interactions, and to tend that companies operating in stable environ- influence SD processes (Armstrong, 1982; Miller, ments follow rational-comprehensive strategic 1987). The literature is replete with studies arguDMPs. In the same vein, Stein (1980) argues ing that FPSs are essential tools for managers, that companies operating in highly dynamic since they are designed to improve managerial environments may tend to employ both less decision-making (e.g., Duncan, 1990; Langley, extensive search and less explicit analysis of 1988). But there is an opposite line of argument, alternatives. Yet, Bourgeois and Eisenhardt which discounts their contribution to SDs. It has (1988) concluded that in high-velocity environ- been convincingly argued that much of the actualStrat. Mgmt. J., Vol. 19, 115-147 (1998) ? 1998 John Wiley & Sons, Ltd.conclude that 'leadership does not constitute a meaningful contextual domain influencing strategic procedures' (Stein, ). The same view has also been supported by Lieberson and O'Connor (1972), and Hannan and Freeman (1977). From another perspective, Lyles and Mitroff (), note that management characteristics may not influence the organizational problem-formulation process. This is a significant issue that needs to be resolved empirically. The influence of top management on SDs remains unclear. To advance our knowledge of the role of the CEO and the TMT we need a better understanding of their impact (if any) on strategic DMPs and/or the underlying characteristics which are important (Rajagopalan et al., 1997; Smith et al., 1994).Strategic Decision-making Processes decision-makingmake take place outside FPSs (e.g., King, 1983;Sinha, 1990). It seems therefore a fruitfulresearchquestion,to exploreempirically this link between formal planning systems and strategicDMPs.119organizationaldecision-making.However, it is worth mentioning that Dean and Sharfman (1993a) as well as the Bradfordstudies (e.g., Hickson et al., 1986) found no differences in to strategicDMPs which could be attributed size.Performance. Since virtuallyall strategicinitia- Corporate control. Several studies have protives require resources, a 'resourceperspective' vided evidence on the importantimplicationsof of may be added to the determinants SD proc- corporate control in strategic DMPs (e.g., esses (Bourgeois, 1981; Pfeffer and Salancik, Lioukas,Bourantas, and Papadakis,1993; Mintzto 1978). Researchrelatingpast performance stra- berg, 1973). The type of ownership or control et much attentegic DMPs is limited (Rajagopalan al., 1993). type is a variablewhich has attracted Much research investigates performance in tion, especially lately in the literature markets on relationto the content of strategy,planning,and for corporatecontrol and privatization.If it is strategy formulationprocesses, ratherthan SDs. hypothesized that nationally owned enterprises An exceptionis the study conductedby Fredrick- display a national style of management and son (1985), who foundthat past performance had national 'culture'in decision-making, while suba negative effect on the comprehensiveness of sidiaries of multinationals may represent an strategicDMPs. implanted (probably more 'sophisticated') More thanthreedecadesago, Cyertand March decision-making style, then it will be of interest (1963) reached the same conclusion, i.e., that to test whether important differences can be have superior performanceis expected to lower the detected.The Bradford groupof researchers intensity with which organizationswill 'search' provided evidence of the existence of different for and analyze information.In the same vein decision-making patterns between British and Bourgeois (1981) and March and Simon (1958) multinational companies operating in Britain suggestedthat slack resourcesoffer organizations (Mallory et al., 1983). Moreover, as suggested the 'luxury' of 'satisficing', and suboptimal by several authors,public vs. private ownership The above arguments lead us to may decisively affect decision-makingpractices decision-making. hypothesize that performance may be negatively and processes (e.g., Lioukas et al., 1993). relatedto rationaldecision-making. Against this, others have found empiricalsup- Towards an integrated research framework port for a positive relationship.For example, Smith et al. (1988) found that, for both small It is evident from the above brief review that: and larger firms, comprehensive outperformed (1) therehas been little researchon the influence less comprehensivedecision-making,and Jones, of broader context on SDs; (2) most of the Jacobs, and Van't Spijker (1992) reportedcon- studies focus on a limited numberof antecedents sistently positive relationshipsbetween organi- while ignoring other importantsources of influin ence on strategic decision-making processes zational effectiveness and comprehensiveness Takentogether,empiricalresults (model underspecification); most of the stud(3) decision-making. in this area are conflicting.This may be due to ies focus on just one characteristic the process of the moderatingeffect of other omitted variables (i.e., comprehensiveness, politics, decentralor (e.g., environment) to model underspecification ization), despite the fact that strategicDMPs are in which characterizes much of the research mul(4) in addition,much and et of the evidence producedis contradictory far (Rajagopalan al., 1993). from establishinga coherenttheory. Firm size. Companysize is usually considered Therefore,we are not able to answerthe quesin to be of importance the context of SDs. Again, tion 'what are the key influences on the process the evidence is far from clear or generalizable. of making SDs?' Is it the external environment Fredricksonand Iaquinto (1989) reported that as the populationecologists would argue, or is it (CEO and top management larger size is associatedwith comprehensiveness the top management in strategic decision-making.Child (1972) also team (TMT)) as the proponentsof management suggested that size affects the framework of choice theories would contend? Do internal? 1998 John Wiley & Sons, Ltd.Strat. Mgmt. J., Vol. 19, 115-147 (1998)120V. M. Papadakis, S. Lioukas and D. Chambers(2) semistructured interviews w (3) completion of two different questionnaires: one general for the CEO and one decisionspecific for the key participant(s); (4) supplementary data from archival sources (e.g., internal documents, reports, minutes of meetings). The research covers 70 SDs in 38 manufacturing firms in Greece. A sequence of steps was followed in order to secure the reliability of data based on participant recall. The process is described in Appendix 1. The sampling frame comprised all manufacturingenterprises in Greece with more than 300 employees, drawn from three industrial sectors (food, chemicals, and textiles)a total population of 89 companies of which 38 participated in the research. The average size of the companies in the sample is 730 full-time employees. In most cases two SDs were studied in each firm, resulting in a sample of 70 SDs. The response rate achieved (approximately 43%) is very high considering the intrusive nature of the research and the fact that top management was asked to devote several hours of its time. Comparison between respondent and nonrespondent firms on the basis of three objective measures (number of employees, total assets, and return on assets), verified the representativeness of the final sample. Reliability and validity considerations A study based on participant recall, though the dominant method of studying decision-making processes, may have inherent limitations (Bouchard, 1976; Huber and Power, 1985; Kumar, Ster, and Anderson, 1993). A number of procedures have been suggested to help reduce their impact, including the use of multiple informants (Kumar et al., 1993). Even these methodologies do not guarantee objectivity. The nature of the present research (in-depth study of one or two SDs in each company, a separate CEO interview, use of archival data), the specific features of the sample (i.e., medium-sized enterprises, existence of few key informants in each SD), as well as the effort required to find informants to discuss in depth often delicate matters, relating to an SD, made it difficult to use multiple informants per SD and to aggregate their responses. Several tactics were followed in an attempt to alleviate possible biases (Bourgeois and Eisenhardt, 1988; Huber and Power, 1985; Kumar et? 1998 John Wiley & Sons, Ltd.enterprise characteristics affect the process? Do different decision-specific characteristics, as perceived by management, lead to different treatment of the decision? Does past performance play any role in influencing the making of SDs? What is the role and significance of formal planning systems? Since these questions remain largely unanswered (Bateman and Zeithaml, 1989; Bryson and Bromiley, 1993; Pettigrew, 1990; Rajagopalan et al., ; Schneider and De Meyer, 1991), it seems that what is needed is an exploratory approach which views the process of making SDs as subject to multiple influences, and examines the effects of factors in three contextual domains: decision-specific characteristics, top management, and context. The present paper addresses these issues by formulating an integrative model of contextual influence on strategic DMPs. The dimensions of the strategic DMP are shown on the right-hand side of Figure 1. The decision-specific characteristics are depicted on the left-hand side. The top block of the diagram indicates management factors while the lowest block indicates broader contextual factors (corporate environment and internal firm characteristics). This these study operationalizes dimensions/factors, and tests their effect on the DMP. The components of the model, together with operationalization and measurement issues, follow the discussion of our methodology. The exploratory nature of the paper should again be stressed. Given that previous studies have reached widely conflicting conclusions the paper aims to provide evidence as to which domains pertaining to the SD process are more important, and which factors within each domain actually influence various dimensions of the process. Further research will be needed to advance and test particular hypotheses. RESEARCH METHODOLOGYData collection and sampling issues To achieve these objectives an ambitious study was designed and executed, which took more than 14 months of intensive fieldwork. This can be characterized as a multimethod, in-depth field research study (Snow and Thomas, 1994). The data sources include: (1) initial CEOStrat. Mgmt. J., Vol. 19, 115-147 (1998)0.---------- ....... ..........* * * *......... .....C E O Risk Propensity Education Need for Achievement A%gressiveness----------M AN A GEME TOP * Level of Education * Aggressive Philosophy................. ................. ....................._'Norpp C. A*GENERIC CHARACTERISTICS (e.g.) =:& Magnitude of Impact=* * * *Rationality/comprehensiveness Financial Reporting Formalization Hierarchical Decentralization Lateral Communication Politicization Problem solving Dissension& Threat/Crisis Frequency*TYPE OF SD (e.g.) =& New Business Investment =& Investment in Capital Equipment & Investment in Marketing* * *................................. ...-.-. . ...... .....-R ------.-K-4---f-R. .. .3I, -Q~O BEXTERNAL ENVIRONMENT * Heterogeneity * Dynamism * Hostility............ ...........:................:-:.............. . ..--..4INTERNAL CONTEXT *? Internal Firm Characteristics (e.g. planning formality) *? Performance (return on assets, growth) * Corporate Control/Ownership Type (e.g. SOE, private Greek Comp * Size (nr of employees) Figure 1. Factors influencing strategic decision-making processesto122V. M. Papadakis, S. Lioukas and D. Chambersafford some confidence that common method bias was not a problem. Finally, the willingness and sincerity with which top managers participated in the research and the interest they showed during the interviewing process provide a further reason to believe in the face validity of their responses.al., 1993). First, archival records documenting the process and its characteristics were collected prior to each main interview. Second, all the discussions were recorded. This tactic enabled the researcher to have direct access to the original discussion and pay attention to any part of it, at later stages. Interview notes, impressions, and noteworthy points were written down during the first 24 hours after the completion of the interview (Bourgeois and Eisenhardt, 1988). These improved knowledge of the process. Third, particular caution was exercised to minimize distortion and memory failure problems notably by selecting recently taken decisions (Mintzberg et al., 1976), by interviewing only major participants having an intimate knowledge of the process (Kumar et al., 1993), by adopting a 'funnel sequence' method in conducting interviews (Bouchard, 1976), by cross-checking interview-derived information against other managers' recollections (e.g., CEOs), by using additional informants in cases of incomplete information, and by cross-checking interview data with other company sources available (e.g., documents, reports, minutes of meetings). In addition, a small number of key process variables were measured independently (e.g., planning formality, internal reporting activities) based on archival data. Statistical tests showed that managers' recollections were significantly correlated with the selected archival data. In addition, both subjective and objective data on corporate performance were obtained. The two methods provided similar results, reinforcing belief in the validity of the data. However, managers' recollections were used in measuring most of the variables in this study. Another major consideration was the minimization of common method bias. To correct for such effects the following precautions were taken. First, a number of variables (e.g., size, performance) are archival, obviating any danger of common method bias with them. Second, two different questionnaires (general and decisionspecific) were used and they were answered by different managers (i.e., dependent and independent variables were answered by different persons). Third, the items used in the analysis were distributed throughout a lengthy interview. Fourth, scale anchors were reversed in several places to reduce and compensate for the development of response patterns. These precautionsStrat. Mgmt.J., Vol. 19, 115-147 (1998)OPERATIONALIZATION AND MEASUREMENT OF SD PROCESSESThe dimensions along which SD processes were measured in this study are: comprehensiveness/reality, extent of financial reporting, existence of a set of formalized rules guiding the process, hierarchical decentralization, lateral communication, politicization, and problemsolving dissension. These are indicated in Appendix 2 along with their measurement details, reliability levels, and the sources from which they were derived. As outlined in the theoretical framework, ample theoretical support can be found for the above dimensions. For example, the framework adopted is similar to that of Cray et al. (1988). Indeed, the scrutiny dimension is captured by the comprehensiveness and financial reporting dimensions, the interaction dimension is captured by the politicization and problem-solving dissension dimensions, and the centrality dimension is similar to our hierarchical decentralization and lateral communication dimensions. Moreover, the reliability levels are very satisfactory. Especially for the comprehensiveness/ rationality construct, they are higher than those reported by other researchers (e.g., Dean and Sharfman, 1993b; Fredrickson, 1984; Smith et al., 1988). It is noteworthy that despite the fact that the resulting variables tap dimensions of the same phenomenon (i.e., the strategic DMP) they do not have very high intercorrelationcoefficients (see Table 1). All, however, are in the expected direction. For example, the formalization construct is positively and significantly related to the notion of rationality, an association argued by several researchers (e.g., Langley, 1989).? 1998 John Wiley & Sons, Ltd.Strategic Decision-making Table 1. Correlations among SD process dimensions Mean S.D. 1. 2. 3. 4. 5. 6. 7. Comprehensiveness/rationality Extentof financialreporting activities Rule formalization Hierarchical decentralization Lateralcommunication Politicization dissension Problem-solving 3.26 0.730.00a 1.000.OOaProcesses1231 1.00 0.46*** 0.28** 0.42*** 0.62*** 0.31** -0.012345672.77 2.23 2.97 2.501.00 0.54 0.56 1.34 1.271.00 1.00 0.19 0.35** 0.01 1.00 0.35*** 0.27* 0.54*** 1.00 -0.11 0.14 0.10 0.07 1.00 -0.14 0.01 -0.12 -0.08 0.27* 1.00aVariables marked with an asterisk are factors (principal components) incorporated in the analysis.at at at *Significant 0.05; **Significant 0.01;***Significant 0.001.SELECTION AND OPERATIONALIZATION OF EXPLANATORY VARIABLES Measuring decision-specificcharacteristicsTo derive generic dimensions, the present research specified and measured 16 decisionspecific characteristics, which would apply across the diverse SDs in our sample, and which were based on the literaturereview. These issue characteristics are shown in Table 2. These initial variables were factor analyzed, using varimax rotation method, and six factors were derived. Table 2 presents the results of the factor analysis investigation. It is worth noting that all factors reflect distinct, internally consistent patterns suggesting generic characterizations of SDs. A specific name is assigned to each factor based on the variables loading. The names of these factors are: SD's magnitude of impact, uncertainty, amount of pressure anticipated by the participants, frequency/familiarity, extent to which the SD was perceived as a crisis situation, and finally extent to which the SD emerged through the formal planning system (planned vs. ad hoc). Appendix 3 presents details on variable measurement, sources in the literature from which these were drawn, and their reliability levels. But these generic characteristics may not cover the true nature of a project. So further objective decision-specific constants were added describing any idiosyncratic aspects of SDs not accounted by the characteristics included (see Hickson et al., 1986; Shirley, 1982). For the purposes of the present paper, a fourfold classification of SDs is identitied: new business investment decisions (e.g., acquisitions, mergers, joint ventures, new company establishment), investments in capital? 1998 John Wiley & Sons, Ltd.equipment (e.g., expansion of production equipment, storing facilities, modernization of production equipment), investment in the marketing domain (e.g., new product introduction, marketing channels), and finally internal reorganization investments (e.g., investments in information systems, internal reorganization). All these are measured using dummy (0/1) variables. Measuring top management characteristics Both personality and demographic variables are used to measure characteristics of the CEO and the TMT. This would help find out whether it is the CEO, or the TMT, or both, that play an important role in the making of strategic decisions. CEO's personality and demographic characteristics Two CEO personality characteristics are incorporated in the present work: need for achievement, and risk attitude. Need for achievement is, according to several writers, one of the basic characteristics positively associated with entrepreneurial success (Gough, 1976). In the present study Steers and Braunstein's (1976) scale is used (see Appendix 3). Attitude towards risk (risk propensity) is a psychological disposition of individuals to show varying degrees of risk-taking or risk avoidance behavior. It is among the major personality dimensions which was found to be associated with various strategic configurations. The particular construct used is derived from Jackson (1976), and Eysenck and Wilson (1975). Appendix 3 describes how these dimensions were operationalized and measured. The resultingStrat. Mgmt. J., Vol. 19, 115-147 (1998)t :ic,??\oZETable 2.Factor analysis results of decision specific characteristics Factor loadingsa Factor 1: Magnitude of impact 0.805 0.782 0.760 0.743 0.608 0.731 0.726 0.647 Factor 2: Uncertainty Factor 3: Threat/crisis Factor 4: PressureDecision-specific characteristics SDs' impact on strategy Radicality of changes SDs' magnitude of impact Precursiveness of the SD Seriousness of consequences Action uncertainty Overall uncertain nature of SD Information uncertainty Extent of crisis perception Perceived threat of loss Time pressure to make the SD Pressure on the organization SD as part of another decision Frequency of occurrence Familiarity with the SD 'Planned' vs. 'ad hoc' Eigenvalue Percentage of variance explained Cumulative percentageFact Freq fami0-0.266 0.847 0.8190 0.903 0.7710.313 -0.401 -0.282 -0.437 -0.263 3.59 22.4 22.4 2.42 15.1 37.6 1.87 11.7 49.30.3890 0 0zoo 00I. 1?1.38 8.6 57.91 6 64.&Alpha factoring method was used, together with varimax rotation and Kaiser normalization. Factor loadings less than 0.25 arecnr, p-reliabilitycoefficients are satisfactory,providing in reliabilitylevels similarto those reported other Internalsystems. For the purposesof the present studies (e.g., Budner, 1962; Steers and work, planning systems are used as potentially Braunstein,1976). very relevantto SDs (e.g., Sinha, 1990). In partiAs regardsCEO's demographic characteristics cular the variable formalization of the planning several variableshave been used for describing effort is used. It has been suggested by various elites'. The present researchers characteristics 'managerial of (e.g., Grinyer,Al-Bazzaz, and Yasairesearchuses two variables:(1) CEO's length of Ardekani,1986) that formalization one of the is service in the company (number of years with most prominentcharacteristics planning sysof the company);and (2) CEO's level of education. tems. The specific constructused is adaptedfrom Both variables have been reported to have a the work of Wood and LaForge (1981). Only profound influence on organizationalprocesses seven of the initial 18 dimensions proposed by and outcomes (e.g., Finkelstein and Hambrick, Wood and LaForge (1981) have been selected. These dimensions were preferredbecause they 1990; Hitt and Tyler, 1991). refer to the long-termplanningconductedrather than to short-term budgeting practices (see TMT characteristics Appendix 3). When factor analyzed the seven Two measureswere used. The first measuresthe dimensionsproduceonly one factor, furtherverof degree of aggressiveness of what Hage and ifying the appropriateness the modified scale Dewar (1973) call the 'behavioralelite group' used. in (i.e., the CEO and all those participating major decisions). It drawsfrom Khandwalla (1977) and Corporate performance. Two objective measStein (1980), and is measured by three items ures of performanceare used: first, return on expressing dimensions of the TMT's attitude assets (ROA), which is viewed as an operational The firstitem mea- measureof the efficiency of a firm with regard towardsrisk and achievement. suresthe degreeof 'beat-the-competition' attitude, to the profitable use of its total asset base the second TMT's risk propensity(i.e., attitude (Bourgeois, 1980); second, growth in profits, towards risky projects), and the third the top indicatingthe trendin profitability improvement. team's attitudeto innovation.The combination as of This paper treatsperformance an independent these three items is explainedas TMT's aggress- variableinfluencingthe strategicDMP. To assure measureswere calculatedgoing iveness towards competitors, innovation, and this, performance 5 years prior to the decision studied. This adds risky projects. The second variable attempts to capture the confidence in testing whether past performance when makingthe SD. level of education of what Hage and Dewar was a seriousconsideration (1973) name as formal elite. It is an objective variable measuringthe percentageof managers, Firm size. To measuresize this paper uses the down to the level of departmental heads, who log of full-time employees (e.g., Fredrickson, are universitygraduates. 1984). Corporatecontrol. Finally, to capturethe effect of type of ownershipcontrol on decision-making practices two dummy (0/1) variables are used, Environmental context state-owned (SOEs) and enterprises distinguishing Three environmentaldimensions are measured private Greek companies from subsidiaries of using perceptionsof top managers:(1) environ- multinationals. Broader context (2) en and (iii) environmental hostility (oppositeStrategic Decision-making Processes Internalcontext125to munificence). Appendix3 describeshow these dimensions were operationalizedand measured. Cronbachalpha reliability coefficients are satisfactory, providing reliability levels similar to those reportedin other studies which used the same measures.? 1998 John Wiley & Sons, Ltd.DATA ANALYSIS CONSIDERATIONS Given the number of variables involved, both dependent and independent,separate regression models were appliedfor each SD dimension.TheStrat. Mgmt. J., Vol. 19, 115-147 (1998)126V. M. Papadakis, S. Lioukas and D. Chambers ordinary change in regression coefficients. Most of the regression coefficients appear to produce algebraic signs according to theoretical expectations and the coefficients for each regression show a high consistency with single correlations.results reported here present the 'best' regression equation, i.e., the equation which provides the maximum number of significant variables. These would give a first indication of the relative influence of the explanatory variables on each independent variable. A consideration in presenting the models was whether a full equation model should be presented along with the best equation model, for each of the dependent variables. The size of the sample (i.e., 70 SDs) theoretically would impose limits on the number of variables to be introduced simultaneously. The degrees of freedom would not be adequate to assure reliable and valid results in a full variables version. Moreover, the research is exploratory and in most cases there are not prior strong reasons to expect relationships with all variables. Versions with full equations can be obtained from the first author upon request. These do not change the pattern emerging. Each model was derived by both backward elimination and stepwise regression methods in corroboration. In most cases the results were identical. In the very few cases where the two methods provided different equations, further tests were attempted by entering and removing variables from the equation, and finally the model with the best explanatory power was selected. A second consideration refers to possible multicollinearity effects. Table 3 presents the intercorrelations between the independent variables. Only two out of over 300 single correlation coefficients are above 0.50, and then only slightly, indicating that intercorrelations are not unduly high. To safeguard for multicollinearity effects, the procedures outlined by Belsley, Kuh, and Welsch (1980) were followed. According to this, condition indexes were calculated for each of the regression models of Table 4. These condition indices were far below the suggested warning level of 10.0 for mild collinearity. Thus, no serious problems are expected (Belsley et al., 1980). Several other warning signals were also inspected, in order to detect possible multicollinearity problems. In none of the equations is there a substantial R2 accompanied by statistically insignificant coefficients, to raise suspicion about possible multicollinearity problems. The stability of regression coefficients was also tested. Here, several runs were conducted by dropping or adding independent variables in the equation. None of these trial runs has indicated any extraStrat. Mgmt. J., Vol. 19, 115-147 (1998)RESULTSTable 4 summarizes the results of the paper. All models afford good to satisfactory predictions of the extent to which each SD process dimension is determined by the decision-specific, management, and contextual characteristics. The explanatory power of the models ranges from 0.39 to 0.63 and on average exceeds 0.50. Considering the cross-sectional nature of the research, and comparing the results to similar efforts, this is seen as very satisfactory (e.g., Dean and Sharfman, 1993a; Stein, 1980). Comprehensiveness/rationality Overall, results show that comprehensiveness/ rationality is affected by decision-specific characteristics and internal context. Environment and management factors are insignificant. The specific coefficients of Model 1, Table 4 suggest that SD's magnitude of impact and type of SD are the most important decision-specific characteristics while planning formality, corporate performance, firm size, and ownership/control type are the dimensions of the internal context which significantly influence comprehensiveness in the SD-making process. Results are in line with Dean and Sharfman (1993a) and Stein (1980), who suggest that the perceived magnitude of impact of a decision is among the strongest explanatory variables of decision-making behavior, as decision-makers act more comprehensively/rationally when decisions imply important consequences. It is also noteworthy that SDs for new business investment and marketing type seem to be subject to less comprehensive/rational analysis than SDs on capital investment and internal reorganization. This follows from the negative coefficients of the respective dummies. As regards internal context, all dimensions appear to be significant. Results support the normative view that formal planning systems (FPSs) contribute to more rational decision-making? 1998 John Wiley & Sons, Ltd.0 :rr F, 0 0&0~cn 0 Y,Table 3. Intercorrelationsamong independent variables Variables 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. Magnitude of impact Thread/crisis Decision uncertainty Frequency Pressure 'Planned vs. ad hoc' Heterogeneity Dynamism Hostility Planned formality CEO's need for achievement CEO's risk propensity CEO's no. of years with company CEO's level of education TMT's level of education TMT's aggressive philosophy Return on assets Growth in profits Size State-controlledenterprises Private Greek companies Subsidiaries of multinationals Mean 3.29 3.04 2.44 2.97 4.44 3.81 2.71 3.17 2.94 3.04 2.15 3.14 18.14 3.36 85.0 3.35 0.997 1.29 2.76 0.19 0.47 0.34 S.D. 0.75 1.26 0.98 0.80 1.53 1.08 0.93 0.87 0.74 1.13 0.63 0.56 10.3 0.89 16.6 1.18 3.39 5.38 0.27 0.39 0.50 0.48 1 2 3 4 5 6 7 8 9 10 11 12 131 1 09 1 -16 01 1 03 13 -26 1 33 30 03 05 1 42 10 -30 20 13 1 13 -13 -04 -19 19 35 1 23 -25 -26 09 16 21 44 1 -19 05 38 -09 -04-19 -17 -32 1 26 19 -04 17 17 47 31 21 -28 1 -01 -26 00-18 -13 03 47 36 02 -05 1 24 15 -16 -01 39 14 24 36 -09 21 -01 1 -01 -18 00 25 -17 -02 -13 -25 -12 17 -07 -26 00-01-01 10 02 01 -12 30 07 11 05 17 -39 08 25 -31 18 29 27 07 10 -46 44 -43 19 00 37 -13 -13 13 26 34 42 55 -42 53 12 44 01 08 23 03 -04 11 18 13 10 13 16 01 28-14 03 03 -33 -07 22 17 14 17 -33 16 10 10-06 -05 -19 00 00 11 -21 06 -22 16 10-06-04-07 8 -19 -05 -12 09 -18 -08 -02 -23 -17 25 12 -26 -01 -14 07 08 -27 -21 -30 -28 -33 -10 -04 06 -03 13 22 38 35 39 -42 47 16 05 -03 15 -06-17coefficientswere omitted. Decimalsof correlation thanr & 0.275 p & 0.01; For coefficients greater thanr & 0.20 p & 0.05; For coefficientsgreaterthan r & 0.35 p & 0.001. For coefficients greaterocr2.Table 4.0Summary table of best models of regression analysesaModel 1: Comprehensiveness/ rationality Model 2: Financial reporting Model 3: Formalized rules Model 4: Hierarchical decentralizationVariables70Model 5: Lateral communicaoJ\o\oDecision-specific characteristics A. Generic characteristics 1. Magnitudeof impact 2. Threat/crisis 3. Decision uncertainty 4. Frequency 5. Pressure 6. 'Planned' vs. 'ad hoc' B. Type of SD 7. New business investmenttype 8. Investmentin capital equipment 9. Investmentin marketing 10. Investmentin internalreorganization Top management characteristics 1. CEO's need for achievement 2. CEO's risk propensity 3. CEO's numberof years with the company 4. CEO's level of education 5. TMT's level of education 6. TMT's aggressive philosophy Broader context A. Externalcorporateenvironment 1. Environmental heterogeneity 2. Environmental dynamism 3. Environmental hostility B. Internalcontext Internal firm characteristics 1. Planning formality Corporate performance 1. Returnon assets 2. Growth in profits Firm size (no. of employees) Corporate control 1. State-controlled enterprises(SOEs) 2. Private Greek companies 3. Subsidiariesof multinationalsR20.35**0.19* 0.15t -0.16t 0.25*-0.20t -0.19*0.41*** 0.20* -0.25*0.48**0.13t -0.19*0.25* 0.42* 0.64*** 0.37*-0.27** -0.19* 0.20t 0.19t0.29** 0.23** 0.21t 0.35**0.34**0.29** 0.20* 0.19* 0.18t -0.25* 0.63 0.58 11.6*** 0.60 0.54 10*** 0.48 0.43 10*** 0.54 0.489***0.25** 0.29** 0.33***O-s0 0ot~rCLAdjusted R2 F0.66 0.63 24.4***Strategic Decision-making Processes(Armstrong, 1982; Duncan, 1990; Langley, 1988). Contrary to the line of reasoning which understates the contribution of FPSs to decision-making (e.g., King,1983; Sinha, 1990), here planning formality appears to be an important contributor to comprehensiveness in SD-making. A positive relationship between corporate performance and comprehensiveness/rationality is obtained with return on assets. It appears that high levels of performance may produce enough resources to help in making better, mor that may mean that high performers are 'offered the luxury' to invest in more analysis while poor performers may lack these slack resources. The opposite may also obtain: high rationality may lead to better performance thus reinforcing a positive relationship. In the same vein, others have argued that more rational decisions may themselves lead to better performance (Grinyer and Norbum, 1977-78; Smith et al., 1988). In our case this explanation is less likely, given that ROA figures used concern previous years. The results seem to contradict the opposite school of thought, which suggests that superior performance may lower the extent to which organizations engage in rational decisionmaking (Bourgeois, 1981; Cyert and March, 1963; Fredrickson, 1985). Size constitutes a significant explanator of comprehensiveness. This in line with the suggestions of previous work (e.g., Fredrickson, 1984; Fredrickson and Iaquinto, 1989; Mintzberg, 1973), which state size that entails more decision behavior. comprehensive/rational Finally, both dummy variables measuring control/ownership appear to be significantly related to decision comprehensiveness/rationality. More specifically, using subsidiaries of multias nationals benchmarks, state-controlled seem to follow more rational procenterprises esses, while enterprises of private Greek ownership appear to be less rational. This is an interesting result since it verifies the popular view that Greek private enterprises follow less comprehensive/rational DMPs when making decisions of a strategic nature, in comparison to subsidiaries of multinationals. That SOEs are closer to multinationals in this respect is unexpected. Maybe they have qualified staff and engage in more analysis, taking all the necessary time and effort to collect all necessary information and explore alternatives, in order to justify their final? 1998 John Wiley & Sons, Ltd.129choices in their strong stakeholders, both inside and outside the organization. The insignificant coefficients also provide useful suggestions. A striking result is the lack of significance of top management characteristics, since no variable loads significantly in Model 1. Only TMT's aggressive philosophy is marginally significant (at a 10% level). This seems at odds with current theory, which stresses the vital role of the CEO and/or the TMT on comprehensive/ rational decision-making (Hambrick and Mason, 1984). A similar observation can be made as regards environmental variables in determining comprehensiveness, which runs against propositions of environmental determinism. We would expect environmental variables to influence rationality/ comprehensiveness. Indeed, several researchers have argued for the significance of environmental heterogeneity in determining strategic processes (Lindsay and Rue, 1980; Smart and Vertinsky, 1984). These authors argued that managers who perceive their corporate environment as complex tend to emply more comprehensive strategies. Bourgeois and Eisenhardt (1988), in their study of strategic decision processes in high-velocity environments, concluded that effective firms follow the rational model in decision-making. Others have argued that companies operating in stable environments follow rationalcomprehensive processes in making and integrating strategic decisions (e.g., Cyert and March, 1963; Fredrickson, 1984; Fredrickson and Iaquinto, 1989; Hart, 1992). The argument behind this contention is that strategists usually find it difficult to rely on formal financial analysis, indepth study, and rational processes when having to deal with unstable, high-velocity environments characterized by information scarcity and rapid change. Instead, they are obliged to take quick, bold decisions in many instances, relying on the available amount of information. In the same vein, companies operating in stable environments rarely face significant opportunities and thus when having to deal with such a situation they employ more rational processes. Our results do not support either line of thought. On the contrary, they are in line with Smith et al. (1988), who reported a lack of any statistically significant relationship between environmental dimensions and rationality of strategic decision-making. However, one should conStrat. Mgmt. J., Vol. 19, 115-147 (1998)130V. M. Papadakis, S. Lioukas and D. Chamberssider the results of the other dependent variables, educated CEO is thus likely to demand more before rushing to conclusions. detailed information, leading to more financial reporting (Bantel, 1993). All other top management characteristics (e.g., CEO's need for Financial reporting achievement, risk propensity or tenure) appear to Financial reporting, a dimension of rationality be insignificant. From internal firm characteristics, again pertypically applied in SD especially of an investment nature, is significantly affected by decision- formance in terms of ROA is positively associated specific characteristics, and some characteristics with financial reporting. An important finding is of the internal context. also the negative association between private More specifically, from a quick inspection of Greek ownership and financial reporting. It Model 2 in Table 4 it appears that two decision- implies that Greek private firms may rely less on specific characteristics, notably magnitude of formal financial reporting activities when making impact and emergence of the SD through plan- strategic decisions than multinationals. This ning, CEO's level of education and return on strengthens our argumentation when discussing assets are to be positively associated with finan- the results of the comprehensiveness/rationality cial reporting, while private Greek ownership has model. SOEs are not different from multia negative association. Of note are the marginally nationals. of associations level Of note is the general lack of significnce of (10% significant two other generic external corporate environment. This is in line significance) provided by with the results obtained the decision-specific characteristics. using results indicate that situations perceived rationality/comprehensiveness construct. InterestFirst, as crises are actually associated with more finan- ingly, size does not seem to be significantly cial reporting activities. This is in line with pre- associated with financial reporting. These findings vious theoretical argumentation (Dutton, 1986). will be further discussed in later sections of In general, one might argue that when adversity the paper. looms everyone might want to interpret and explain the situation in terms of financial analysis Rule formalization and reporting. Or, taking another view, the company may seek to exercise control and support Rule formalization in the SD process (Model the meaningfulness of its actions in the eyes of 3, Table 4) is influenced by decision-specific both internal and external stakeholders by relying characteristics (decision uncertainty and emeron deeper financial reporting and analysis, since gence through formal planning), top management crises usually involve risks of a significant finan- characteristics (CEO risk propensity) and corpocial loss. rate control type (Greek ownership). financial reporting is negatively related From decision-specific characteristics, decision Second, to frequency. This result supports the view that uncertainty and emergence are negatively frequent/familiar issues are dealt with by standard associated with rule formalization of SD process. rules and analogies from memories. They are Uncertainty, as used here, refers to specific therefore associated with less analysis and com- decisions, as opposed to the uncertainty caused prehensive reporting of data (Marmaras, Lioukas, by the organizational environment. Results are in and Laios, 1992). Also the coefficients of the line with Thompson (), who contends dummies 'investment in capital equipment' and that in cases of high uncertainty managers act in 'investment in marketing' are marginally signifi- an 'inspirational' manner, by making obsolete any cant showing a higher level of financial reporting formal procedures and rules usually followed. for these types of decisions, as against invest- One can contend that high uncertainty about the ments in internal reorganization. decision may, contrary to the received (common) As regards top management, only CEO's level expectations, result in more intuitive processes of education is positively associated with financial (Daft and Lengel, 1986; Dean and Sharfman, reporting. Education level has been found to be 1993a) together with use of less formalized rules. related to the extent of people's information Again, as expected, SDs emerging from the discisearch and analysis (Dollinger, 1984). A highly pline of a formal planning system are found to follow more formalized paths.Strat. Mgmt. J., Vol. 19, 115-147 (1998)? 1998 John Wiley & Sons, Ltd.Strategic Decision-making Processes we As regardsmanagement observe a negativeassociation between CEO's risk propensity and131in rule formalization taking SDs. Again, such a result is intuitively expected, since risk-takers usually break the bounds of organizational systems and formalitiesand influencethe SD process towards more informalpaths. TMT's aggressive philosophy is also related to more formalized rules.Finally, private Greek companies seem to beas lagging in rule formalization the negativecoefficient suggests. SOEs do not differ from multinationals. Overall, however, we cannot argue that there exists a 'balanced contribution'of all domains (SD itself, management, and context) in explaining SD formalization.Of interest is the lack of significantimpact of decision frequency, corporateenvironment,planning formality, past performance,and size. This suggests that SD from the formal process formalityis independent planning machinery of the firm and external environment the otherinternal and comcorporate pany characteristics.It is more a matter of decision-specific characteristicsand top managementchoice. Hierarchical decentralization and lateral communication The extent to which the SD process is decentralized and allows participation lower-levelmanofagers depends on the decision-specific characteristics, on CED tenure, and corporate profitability.is Moreover,the extent of lateralcommunicationdetermined by decision-specific characteristics, top management team's aggressiveness, and internal firm planning formality.More specifically, several conclusions can be drawnfrom the resultsof Models 4 and 5, Table 4. From the decision-specific characteristics, As regards the effect of type of investments on strong is the effect of magnitudeof hierarchical decentralization,all dummy coefparticularly impact, followed by perceived pressure and ficients are significant,showing that only internal threat/crisis. Results imply that SDs with reorganizationinvestments are relatively more importantimpact attractthe collective attention centralized.On the contrary,none of the dummy of more layers in the hierarchyand more depart- variables is significantlyassociated with lateral mentsas revealedin Models4 and 5. This corrob- communication. oratesDuttonet al. (1989), who arguethat issues Top management has two significant coefCEO's with great magnitudeof impactimply high inter- ficients.In Model 4 the variablemeasuring decenconnectednesswith other relevant issues. There- tenure is positively relatedto hierarchical This may be explainedby the fore, such issues attractmore collective attention tralization patterns. decentrali- fact thatCEO's tenuremay influenceparticipation and thus result in higher hierarchical zation and lateralcommunication. patternsby developing greater levels of social? 1998 John Wiley & Sons, Ltd.Pressurehas a negative coefficient, suggesting that when SDs are taken under pressure there may not be enough time to involve more levels and departments. An interestingeffect obtains for threat/crisis. This suggests that threatening situationsresult in more hierarchical decentralization. first this is At A counterintuitive. numberof authors(e.g., Dutton, 1986; Herman, 1963) argue that centralization of authorityis the expected outcome of crises, since two opposite forces clash. First, managerialelites undertakethe responsibilityof the whole effort to divert the crisis. Second, middle managers,feeling that the issue might be 'too heavy' for them to deal with, pass it to top management. Milbur et al. (1983) provide an explanation for this counterintuitive result:their findingssugwas gest that althoughcentralization the immediate outcome of crises, the actual intermediate of response was decentralization authority.This of character may be explainedby the distributed info if we admit that the sourceof vital information middlemanagement, is centralization deprives top management of extremely useful data. Herman (1963) offered anotherexplanation.He arguedthat the relationof ship between crisis and decentralization authThus, under situationswhich ority is curvilinear. are characterized 'mild crises' one may observe as decentralization. contrast,when crises become By acute, authoritycentralizationis found. Further investigatingthe descriptivestatisticsof the variable measuring extent of perceivedcrisis, we the may see that the variable is measuredon a 5point scale and has a mean of 2.51, which implies rather'mild crises', on the average.This suggests that our sample of SDs may not include intense crises which are assumed to lead to hierarchical centralization less lateralcommunication. andStrat. Mgmt.J., Vol. 19, 115-147 (1998)132V. M. Papadakis, S. Lioukas and D. Chambersa climate of shared effort, and facilitates smooth implementation of strategic decisions. Smooth implementation contributes in turn to higher performance (Wooldridge and Floyd, 1990). On the other hand, lack of involvement of employees other than 'strategic elites' in the process has been found to create implementation problems, inclu