University of Southern California Dissertations and Theses

Does adding nonfinancial value drivers to a summary financial measure improve the learning and performance of managers?

(USC Thesis Other)

Does adding nonfinancial value drivers to a summary financial measure improve the learning and performance of managers?

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DOES ADDING NONFINANCIAL VALUE DRIVERS TO A SUMMARY FINANCIAL MEASURE IMPROVE THE LEARNING AND PERFORMANCE OF MANAGERS? Copyright 2003 by Lay Khim Ong A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (BUSINESS ADMINISTRATION) May 2003 Lay Khim Ong R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. UMI Number: 3103953 Copyright 2003 by Ong, Lay Khim All rights reserved. ® UMI UMI Microform 3103953 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90089-1695 This dissertation, written by under the direction o f f£ r dissertation committee, and approved by all its members, has been presented to and accepted by the Director of Graduate and Professional Programs, in partial fulfillment o f the requirements fo r the degree of LAY KHIM O N G DOCTOR OF PHILOSOPHY Director Date Mav 1 6 . 2 0 0 3 Dissertation Committee Chair R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. ACKNOWLEDGEMENTS I want to thank my committee members: Sarah E. Bonner, Mitchell Earleywine, Kenneth A. Merchant (Chair), and Wim A. Van der Stede for their guidance and support. I also want to thank Mihail Bota for programming assistance. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. iii TABLE OF CONTENTS Acknowledgements ii List of Tables iv List of Figures v Abstract vi Chapter 1: Introduction 1 Chapter 2: Theory and Hypotheses 12 Chapter 3: Experimental Design and Procedures 44 Chapter 4: Data Analysis 76 Chapter 5: Conclusions 111 Bibliography 117 Appendix A: Experimental Stimuli 133 Appendix B: Properties of the Natural Log Function 180 Appendix C: Calculations of Optimal R&Dt- 3 , P&Et . 3 and nt 183 Appendix D: Calculations of Expected nt, R&Dt, P&Et, R&DPt, and P&EOEt 185 Appendix E: Calculations of Expected Points Reward for Each Condition 189 Appendix F: Calculations of MSD between Actual and Optimal Investments 200 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. iv LIST OF TABLES Table 1: Classification of Heuristics 32 Table 2: Experimental Design and Descriptive Statistics 46 Table 3: Sample Combinations of R&Dt - 3 and P&Et- 3 , and resultant nt, 57 R&DPt-s and P&EOEt.3 Table 4: Description of Multiple Contrasts 77 Table 5: Critical Two-tailedp -values under Holm’s Test for 10 Multiple 78 Contrasts Table 6: ANOVA for Performance 81 Table 7: Performance (in $ million) by Condition 82 Table 8: ANOVA for Learning 87 Table 9: Learning by Condition 88 Table 10: ANOVA for Optimality of Resource Allocation 97 Table 11: Optimality of Resource Allocation by Condition 98 Table 12: ANCOVA for Performance with Learning as a Covariate 101 Table 13: Performance (Adjusted for Learning) by Condition 102 Table 14: ANCOVA for Performance with Prior Beliefs as Covariates 103 Table 15: ANCOVA for Learning with Prior Beliefs as Covariates 104 Table 16: Performance (Adjusted for Prior Beliefs) by Condition 105 Table 17: Learning (Adjusted for Prior Beliefs) by Condition 105 Table 18: Effect Sizes and Power of Contrasts 109 R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. V LIST OF FIGURES Figure 1: Libby & Luft (1993) model ofjudgment performance 12 Figure 2: Model of determinants of managerial performance 16 Figure 3: Graphical depiction of HI to H4 38 Figure 4: Performance (in $ million) by condition 83 Figure 5: Learning by condition 88 Figure 6: Optimality of resource allocation by condition 98 Figure 7: Performance (adjusted for learning) by condition 102 Figure 8: Performance (adjusted for prior beliefs) by condition 105 Figure 9: Learning (adjusted for prior beliefs) by condition 106 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. vi ABSTRACT Traditional performance measurement systems (PMSs) that use accounting measures have been criticized as inadequate for today’s business environment, especially when intangible assets rather than tangible assets have become the major sources of competitive advantage for firms. To overcome the limitations of accounting-based measures, nonfinancial measures have been recommended because they are believed to be leading indicators of financial performance. With an experiment, this study examines the effects on the learning and performance of managers of adding nonfinancial value drivers (VDs) to a summary financial measure in a PMS. The experiment is designed to separate the effects of information on the performance of VDs (VD information), information on the relative importance of VDs (VD weights), and incentives placed on the performance of VDs (VD rewards). This study also examines whether adding VDs is more beneficial in firms whose competitive advantage is primarily associated with intangible assets rather than tangible assets. The experimental setting is unique in the accounting literature, involving a dynamic multi-period decision making environment where managers make decisions that are interdependent across multiple periods and where financial performance lags decisions. The results indicate that providing VD information and VD weights does not improve managerial performance nor reduce performance variability. Managerial performance improves and performance variability decreases only when VDs are rewarded and only in firms in which intangible assets are more important than R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. tangible assets for future financial performance. In addition, managerial learning does not improve with the provision of VD information, VD weights, or VD rewards. Considering the results for managerial performance and learning together, the improvement in managerial performance is more likely a result of better managerial motivation when VDs are rewarded rather than better managerial learning when VD information or VD weights are provided or when VDs are rewarded. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 1 CHAPTER 1: INTRODUCTION The Importance of Performance Measurement Systems The performance measurement, evaluation, and compensation system (hereafter referred to as performance measurement system, PMS) of an organization is critical to its success. At the organizational level, accurate performance measurement and evaluation is needed to assess how well the organization has achieved its goals, predict its performance in the future, direct management to areas that require attention, and allocate limited resources amongst various business units, products, or activities to best achieve the organization’s objectives. At the individual level, PMSs are important because they affect the incentives and behavior of employees towards increasing the welfare of their organization rather than their personal welfare (Indjejikian, 1999). Empirical evidence shows that the PMSs used in firms affect the behavior of employees and the performance of the firms (e.g., Abowd, 1990; Banker, Lee, & Potter, 1996; Larcker, 1983). The effectiveness of the PMS affects a company’s ability to identify, motivate, and retain the best employees. A PMS can serve the functions of planning goals for job performance for employees, providing feedback on employee’s performance, identifying training and development needs, and determining employees with specific skills and abilities. Companies depend on information from the PMS to make decisions regarding bonuses, pay raises, promotions, and dismissals of employees (DeNisi & Williams, 1988). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 Inadequacy of Financial Measures Traditional PMSs that use accounting measures have been criticized to be inadequate for today’s business environment (American Institute of Certified Public Accountants [AICPA], 1994; Eccles, 1991; Ittner & Larcker, 1998; Johnson & Kaplan, 1987; Seidler, 1986). Accounting measures are lag indicators that capture historical performance but do not predict future firm performance well. Studies examining whether earnings are relevant for predicting the market value of firms, where market value proxies for future firm performance, conclude that earnings are not useful for explaining stock returns with the correlation between earnings and stock returns being very low and unstable (Lev, 1989; Lev & Zarowin, 1999). Based on the findings of returns/earnings studies during the 1980 to 1988 period, Lev (1989) noted that earnings accounted for only 2% to 10% of the variation in stock returns even after varying the return windows and using specific subsamples. Moreover, the retums/eamings relations during the 1982 to 1986 period, in the form of the coefficients of regressions of returns on earnings and the R2, fluctuated considerably. Lev and Zarowin (1999) also showed that the association between earnings and stock returns decreased from 6% to 12% in the 1977 to 1986 period to 4% to 8% in the 1987 to 1996 period, suggesting that the usefulness of earnings in firm valuation declined from 1977 to 1996. Being summary bottom-line measures, accounting measures are too aggregated. As such, accounting measures cannot provide information on the value R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 drivers of the business nor specific guidance to managers on what to do to achieve business goals. Moreover, generally accepted accounting principles (GAAP) used in the production of accounting measures often create distortions in the measurement of economic income and capital base of the business (Ehrbar, 1998; Seidler, 1986; Stewart, 1991). The problem is more acute today as intangible assets rather than tangible assets have become significant sources of competitive advantage for firms (Lev, 2000). Accounting measures can capture investments in tangible assets and the expenses associated with the use of these tangible assets but cannot adequately account for intangible assets (Kaplan & Norton, 2001a, 2001b; Lev, 2000; Lev & Zarowin, 1999). Investments in intangibles such as research and development (R&D), franchise development, customer base and market share development, brands, and human resources, are typically expensed immediately rather than capitalized under GAAP. Such an accounting treatment depresses the earnings and book values of firms with large investments in intangibles, resulting in financial measures that are less relevant for assessing market value. For example, Amir and Lev (1996) found that financial measures such as earnings, book values, and cash flows were not relevant for valuing cellular companies whereas nonfinancial measures such as customer penetration rate and population size, which proxies for growth, were value-relevant. Amir and Lev (1996) argued that cellular companies spent significant amounts of money on intangibles such as customer and market share development that were immediately expensed, thereby decreasing the value- relevance of earnings. Lev and Zarowin (1999) documented that firms that increased R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. 4 their R&D expenditures experienced decreases in the informativeness of their reported earnings. In addition, accounting measures are short-term in nature and may encourage myopic and dysfunctional behavior in managers. Managers may act to increase accounting measures at the expense of long-term performance by reducing R&D expenditure (e.g., Bushee, 1998; Dechow & Sloan, 1991), overemphasizing projects with short-term payoffs (e.g., Merchant, 1990), and delaying discretionary expenditures such as maintenance (e.g., Hopwood, 1972; Merchant, 1990). Earnings management studies also documented that managers often manipulated accounting earnings via accounting procedures and accrual policies to maximize their short-term bonuses based on accounting earnings (e.g., Guidry, Leone, & Rock, 1999; Healy, 1985; Holthausen, Larcker, & Sloan, 1995). Nonfinancial Measures as Improvement to Financial Measures Many suggestions have been made to overcome the limitations of traditional accounting-based measures. In particular, practitioners and researchers have recommended augmenting accounting measures with nonfinancial measures that reflect key value-creating activities, that is, nonfinancial value drivers (AICPA, 1994; Eccles, 1991; Kaplan & Norton, 1992, 1993, 1996). A recent survey of performance measurement practices of 1990 firms from multiple industries such as financial (17.60% of responses), industrial (16.90% of responses), and cyclical consumer (13.80% of responses), with revenues ranging from $20 million to over $10 billion, showed that these firms are using both financial and nonfinancial R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 5 measures although financial measures are still given greater emphasis (AICPA & Maisel, 2001).1 Nonfinancial measures are recommended because they are believed to be leading indicators of financial performance (Kaplan & Norton, 1992, 1996). Managers act and make decisions that affect nonfinancial measures such as innovation, quality, productivity, and customer satisfaction, which ultimately lead to future financial performance. Current financial measures are not likely to reflect the long-term benefits of current actions. Prior empirical research supports the claim that nonfinancial measures are positively associated with future accounting performance (e.g., Anderson, Fomell, & Lehmann, 1994; Banker, Potter, & Srinivasan, 2000; Behn & Riley, 1999; Ittner & Larcker, 1998b). Adding nonfinancial measures is also consistent with theoretical work on performance evaluation using agency theory (Bankar & Datar, 1989; Feltham & Xie, 1994; Holmstrom, 1979). Under the informativeness principle, nonfinancial measures improve the optimal incentive contract if they provide additional information about desirable managerial actions, which are not captured adequately by financial measures (Holmstrom, 1979; Holmstrom & Milgrom, 1991). If managers are rewarded based on financial measures alone, they may pay too little attention to those aspects of managerial performance that financial measures do not 1 The AICPA/Maisel Survey indicated that the performance measures were used for multiple purposes such as measuring business results, managing operations, determining rewards, evaluating individual performance, and supporting decision-making (AICPA & Maisel, 2001). The most commonly used financial measure was revenues (63% of respondents). 61% o f respondents used net operating income and 32% used ROA or ROI. The most commonly used nonfinancial measure was customer R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. 6 assess effectively. Adding nonfinancial measures may then direct managers’ attention to those tasks that would have been underemphasized. Research Questions Despite the managerial interest in nonfinancial measures, there are relatively few research studies that investigate their effectiveness at improving organizational performance (Ittner & Larcker, 1998a). In this study, I focus on the addition of nonfinancial value drivers (hereafter referred to as VDs) to a summary financial measure (hereafter referred to as SFM). VDs in this study reflect the outcomes of actions and activities that create long-term firm value and therefore are leading indicators of future financial performance. Using an experiment, my study addresses the following research questions. 1. Does adding VDs to a SFM improve the learning and performance of managers? 2. Are the effects of adding VDs to a SFM on managerial learning and performance a result of information on the performance of VDs (VD information), information on the relative importance of VDs from the weights on the VDs in the PMS (VD weights), or the incentives placed on the performance of VDs (VD rewards)? 3. Are the effects of adding VDs to a SFM on managers’ learning and performance moderated by the type of firm managed? Specifically, is adding VDs more beneficial in firms whose competitive advantage is primarily associated with satisfaction (70% o f respondents). 47% of respondents used productivity, 40% used quality and R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 7 intangible assets than in firms whose competitive advantage is primarily associated with tangible assets? 4. What are the heuristics that managers use to learn the relationships between their actions and performance? My study fills a gap in existing research on the usage of nonfinancial measures in PMSs. We lack clear and consistent empirical evidence on whether adding nonfinancial measures to traditional accounting measures improves firm performance and on the different contingencies affecting the effectiveness of nonfinancial measures. Empirical evidence indicates that the use of nonfinancial measures increases with the adoption of new manufacturing practices such as TQM, JIT, and manufacturing flexibility (Abemethy & Lillis, 1995; Banker, Potter, & Schroeder, 1993; Ittner & Larcker, 1995, 1997; Perera, Harrison, & Poole, 1997). However, evidence of nonfinancial measures enhancing performance in these situations is not consistent. Also, while the use of nonfinancial measures in compensation contracts has been shown to be contingent upon factors such as strategy, existence of quality programs, level of regulation, length of product development and product life cycle, and growth opportunities, it is not clear whether and when nonfinancial measures improve performance (Bushman, Lidjejikian, & Smith, 1996; Govindarajan & Gupta, 1995; Ittner, Larcker, & Raj an, 1997). My study makes the following contributions. process-related measures, and 22% used innovation and new product development measures. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 8 1. I focus on the use of VDs to strengthen the power of the test of the effects of nonfinancial measures. Archival studies of organizations that have implemented nonfinancial measures may not necessarily be studying measures that reflect value- creating activities and drive future financial performance. 2. I study the performance effects of adding VDs to a SFM at the individual rather than the organizational level. This may help explain the mixed results in archival studies of organizations that have adopted nonfinancial measures. Relatively few studies have examined the cognitive and behavioral effects of adding nonfinancial measures to financial measures at the individual level. Specifically, I examine how adding VDs to a SFM affects the learning and motivation of individuals, which in turn affect their performance in managing businesses over time. 3. I attempt to separate the performance effects of VD information, VD weights, and VD rewards, which are often confounded in archival studies of nonfinancial measures used in organizations. Organizations vary in the way they implement VDs in their PMSs, with some only providing VD information but not rewarding VD performance while others linking VD performance to compensation. Proponents of nonfinancial measures are not clear whether, when, and how these measures should be rewarded (Kaplan & Norton, 1996). Kaplan and Norton (2001a) noted that companies that have implemented the Balanced Scorecard are careful about linking compensation to the measures on the scorecard and firms vary in the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 9 way the measures are linked to rewards.2 When VD performance is rewarded, the weights on the VDs in the incentives scheme may provide information to managers as to the relative importance of different VDs. We are not sure whether it is VD information, VD weights, or VD rewards that drive any performance benefits associated with the introduction of VDs in the PMS. 4. I examine the moderating effect of the organization’s dependence on intangible assets. To my knowledge, my study is the first empirical study to directly examine the learning, motivation, and performance consequences of adding VDs to a SFM across different types of firms. The study of firms with different degrees of reliance on intangible assets is especially timely since there is an increasing number of firms whose success depends more on intangible assets (Lev & Zarowin, 1999). Moreover, proponents of nonfinancial measures often argue that nonfinancial measures are especially pertinent for firms that depend heavily on intangible assets since financial measures are inadequate for these firms (Kaplan & Norton, 2001a, 2001b). 5. I use an experimental setting that is unique in judgment and decision making research in accounting. I examine dynamic decision-making rather than discrete and static decision-making. Dynamic decisions involve a series of decisions rather than one single decision; are interdependent; and are embedded in an 2 The Balanced Scorecard (BSC) is a PMS that combines a “balanced set” of performance measures from four organizational perspectives: Financial, Customer, Internal Business Processes, and Learning and Growth (Kaplan & Norton, 1996). Its proponents argue that the BSC complements traditional SFMs with forward-looking VDs. The performance measures are supposed to reflect the strategy of the organization and capture the critical activities that drive the organization’s long-term success. R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. 10 environment that is affected by both endogenous decisions and exogenous variables (Brehmer, 1990, 1992, 1995; Edwards, 1962). Judgment and decision-making research has tended to overlook the dynamic nature of real decision-making environments, limiting the generalizability of their results (Hogarth, 1981; Kleinmuntz, 1993). I also study judgment performance where subjects are given ample opportunities and monetary incentives to learn. The results in this study indicate that the addition of VDs to a SFM improves the performance of managers and reduces the variability of the performance across managers only under specific conditions. Merely providing VD information or VD weights to managers does not improve performance or reduce performance variability. Performance improves and performance variability declines only in firms that are more reliant on intangible assets when VDs are explicitly rewarded. In other words, VDs benefit firms that are more reliant on intangible assets to a greater extent than firms where tangible assets are more important, but only when VDs are linked to compensation. Moreover, the learning of managers does not improve with the provision of VD information, VD weights, or VD rewards. Considering the results for performance and learning together, the performance improvement is more likely due to better motivation of managers when VDs are rewarded and not from better learning when VD information or VD weights are provided or when VDs are rewarded. The remaining dissertation is organized as follows. Chapter 2 discusses prior literature and the theory behind specific hypotheses. Chapter 3 outlines the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 11 experimental design and procedures, operationalizations of the variables, manipulation checks, and control of threats to internal validity. Chapter 4 discusses the statistical analyses and the results. Chapter 5 summarizes the dissertation findings, discusses the limitations of the study, and provides future research directions. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 12 CHAPTER 2: THEORY AND HYPOTHESES Model of the Determinants of Managerial Performance The effects of adding VDs to a SFM on managerial performance may be mediated by several variables. I focus on two specific mediating variables in this study: learning and motivation. I propose a general model of the determinants of managerial performance before discussing the specific theory underlying my hypotheses. Libby and Luft (1993) proposed a model of judgment performance where knowledge and ability affect performance, and experience and ability affect the acquisition of knowledge (Figure 1). In addition, motivation or effort affects the degree to which people acquire knowledge from experiences and the degree to which people apply available knowledge and abilities to the task. Ability Experience Performance Knowledge Figure 1. Libby & Luft (1993) model of judgment performance I propose a model of managerial performance adapted from the Libby and Luft model (Figure 2). Managers manage different types of firms under different PMSs. Managerial tasks are multi-dimensional and involve many judgments, decisions, and actions that ultimately impact firm performance or the economic R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 13 profits of the firm (hereafter referred to as firm profits).3 In this study, managerial performance is synonymous with firm profits since only the actions of the manager affect firm profits. In reality, many other variables outside the control of the manager may affect firm profits. For tractability, this study focuses on a single managerial task, resource allocation, wherein a manager has to allocate and balance the resources of the firm across two different types of investments. The two types of investments differentially impact firm profits with a time lag. A manager has to learn the time lag or delay between investments and their impact on firm profits as well as the relative importance of different investments for firm profits. Managers learn or acquire knowledge about their tasks through experience and feedback from the PMS (Link 1 in Figure 2). Learning theories generally support the idea that experience and appropriate feedback are required for people to acquire the knowledge necessary for performing a task (e.g., Anderson, 1982; Anzai & Simon, 1979; Chase & Simon, 1973; Langley & Simon, 1981; Larkin, 1981; Lewis & Anderson, 1985).4 The type of firm will moderate the 3 Economic profits are distinct from accounting profits, which are products o f the accounting process. While the performance o f a firm is not necessarily judged only by its economic profits, profits are a top priority for most for-profit firms. Hence, economic profits are set as the sole criterion for firm performance in this study. 4 For example, in Anderson’s ACT system, which identified the two major stages of the development o f a cognitive skill, declarative knowledge is converted to procedural knowledge with practice (Anderson, 1982; Anderson, Greeno, Kline & Neves, 1981). Declarative knowledge is a set of facts about the skill encoded in a prepositional network. Procedural knowledge is a set of procedures for performing the skill encoded as productions. Feedback is necessary for a person to generalize, discriminate, and strengthen productions. A person must receive feedback about whether a production has been correctly applied. Through examples of success and failure, the person is able to generalize productions to novel situations, discriminate and restrict productions to specific situations, and strengthen productions that have been successful. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 14 learning and knowledge acquisition of managers through the PMS (Link 2). Different PMSs are expected to promote different degrees of learning in different types of firms. Managers apply the acquired knowledge to the task, the quality of which will affect the optimality of their resource allocation decisions (Link 3), and thereby their performance (Link 7). Studies show that the quality of learning and knowledge affects performance in a variety of tasks such as software development, chess playing, solving of physics problems, software development, and medical diagnosis (e.g., Adelson & Soloway, 1988; Chase & Simon, 1973; Chi, Feltovich, & Glaser, 1982; Chi, Glaser, & Rees, 1982; Groen & Patel, 1988). The relationship between knowledge and performance has been demonstrated by prior research in various tasks such as ratio analysis in audit planning and identification of errors in financial statements (e.g., Bedard & Biggs, 1991; Bonner & Lewis, 1990; Bonner & Walker, 1994; Libby & Tan, 1994). In addition to its effects on learning, motivation theories suggest that the PMS also directly affects the motivation of managers since it determines the measures on which managers are rewarded (Link 4). Expectancy theory posits that given multiple activities, people would choose and put in the most effort in those activities that maximize the expected value of desired outcomes, one of which may be monetary rewards (Atkinson, 1964; Vroom, 1964). By tying desired monetary rewards to specific measures, the PMS motivates managers to direct resources to increase those measures so as to increase desired monetary rewards. Goal setting theory suggests that goals motivate the actions of people by affecting the amount of R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. 15 effort they put into a task, the persistence of their effort until their goal is achieved, and the direction to which they devote their effort (Locke & Latham, 1990). By tying desired monetary rewards to specific measures, the PMS encourages people to set goals for those measures and commit to those goals (Locke & Latham, 1990, p. ISO- 144; Locke, Shaw, Saari, & Latham, 1981). Moreover, incentives contracts typically stipulate standards of performance that must be achieved before rewards are given. These standards of performance serve as goals that motivate managers. The motivation of managers affects the optimality of their resource allocation decisions (Link 5), which will then affect their performance (Link 7).5 The type of firm moderates the impact of motivation on the optimality of resource allocation (Link 6). If the firm is such that the measures in the PMS are congruent with firm profits, the manager will be motivated to allocate firm resources to maximize both the measures and firm profits. Optimality of resource allocation and managerial performance improve. However, if the firm is such that the measures in the PMS are incongruent with firm profits, the manager will be motivated to allocate resources to maximize those measures rather than firm profits. Optimality of resource allocation and managerial performance suffer. Managerial performance may differ under different PMSs because managers learn differently or because they are motivated differently by different incentives. 5 Bonner, Hastie, Sprinkle, & Young (2000), in a review of 131 studies, found that financial incentives improve performance in about half of the studies. They found that tasks which are cognitively more complex tend to be less sensitive to incentives and that rewards contingent on achievement o f a quota are more likely to improve performance. The complexity of the task and the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 16 I .earning effects should be distinguished from motivational effects. Managers may perform better not because they understand the relationships between investments and firm profits better but because they are investing in a manner that maximizes their monetary rewards. Since both learning and motivation affects the optimality of resource allocation decisions, the experiment is designed to disentangle the effects of learning and motivation. Link 2 Link 3 Link 7 Link 1 Link 4 Link 5 Link 6 PMS Motivation Type of firm Performance Type of firm Resource allocation Learning (Knowledge) Figure 2. Model of determinants of managerial performance The theory behind the individual hypotheses is detailed in the following subsections. Effects of a PMS on Learning The type of PMS affects learning from experience because of the feedback that it provides managers. Feedback is important for learning because it provides information about errors and guidance to managers to correct their responses and structure of the incentive scheme are held constant across all conditions in this experiment and are not R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 17 adjust their mental model of the task (e.g., Gibson, Fichman, & Plaut, 1997; Salmoni, Schmidt, & Walter, 1984; Schmidt, Young, Swinnen, & Shapiro, 1989). People’s mental model of the task guides their judgments and decisions. PMS provides information for managers to test and modify their mental models about how their actions affect the performance of the firm. McKinnon and Bruns (1992) found that managers in their field study used accounting reports to learn the causal relationships between their actions and the success of their firm. Research shows that the mental models that people use for causal attributions are deficient in a dynamic setting with feedback delays and exogenous variables impacting the system (Diehl & Sterman, 1995; Paich & Sterman, 1993; Sterman, 1989a, 1989b, 1994). In judging causation, people use several heuristics such as the temporal and spatial proximity of cause and effect, the temporal precedence of cause before effect, covariation of cause and effect, and similarity of cause and effect (Einhom & Hogarth, 1986). Such heuristics create difficulties in forming accurate causal attributions in systems where cause and effect are separated by time and space, actions have multiple consequences, and simultaneous variables are in action. Due to limited human cognitive capabilities, people lack memory capacity to remember and process all those cause and effect relations (Simon, 1982). Interference from other simultaneous and interspersing events makes attribution of causation difficult. Shanks, Pearson, and Dickinson (1989) argue that in judging causality, people may be using a rule based on the probability that an expected to confound the effects of the independent variables under study. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 18 outcome (O) occurs given an action (A), P(Q/A), and the probability that an outcome (O) occurs without an action (~A), P(0/~A). A delay between an action and an outcome shifts more observations from those used to estimate P(0/A) to P(0/~A), thereby making the correct attribution of the outcome to the action more difficult. Wasserman and Neunaber (1986) suggest that the degree to which a person attributes an outcome to a target action is a direct function of relative contiguity. Relative contiguity is the difference between the delay of an outcome after the target action and the delay of an outcome after other non-target actions. Non-target actions compete with the target action as possible causes for the outcome. Relative contiguity will decrease when the delay between the target action and the outcome increases, leading to deterioration in the judgments of causality. Learning from SFM A SFM is a lag indicator reporting on outcomes from past managerial actions. It is also a summary bottom line measure that does not provide specific guidance to managers on what to do to maximize firm performance. With feedback from a SFM alone, managers have to form their own mental representation of the relationships between investment decisions and firm profits. Since there is time delay between investments and their effects on a SFM as well as random shocks affecting the SFM, a manager’s mental model is expected to be deficient. With a faulty mental model, managers are not able to utilize optimal task strategies and that will impede their performance (Huber, 1995). Specifically, the manager is expected to use a cognitively simpler feedback strategy that does not R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 19 account for time delays as opposed to a feed-forward strategy guided by a predictive model that accounts for time delays (Brehmer, 1990, 1992; Brehmer & Allard, 1991). In a feedback strategy, a person acts to reduce discrepancies between the actual and desired states of the system based on current information about the system, assuming that current information reflects the actual system state. A feedback system works well if there are no significant delays in the system. In a feed-forward strategy, a person uses a model of the system to predict its state and chooses the appropriate actions to bring the system to the desired state. A feed forward strategy can handle time delays by incorporating them into the model of the system. Brehmer (1990) suggests that people are more likely to use feed-forward strategies if the process causing the delays is apparent and does not have to be inferred. Otherwise, people tend to use feedback strategies. Hence, with feedback from a SFM alone, managers may have a deficient mental model and utilize ineffective task strategies that prevent them from recognizing that investments lead firm profits and the type of investments that impact firm profits more. Research also finds that when there are time delays between an action and its outcomes, it creates barriers to learning (Brehmer, 1990; Sterman, 1994). Delays reduce the number of learning cycles. This slows the process of accumulating experience, testing hypotheses, and improving performance. In addition, delays reduce the amount of learning in each cycle. The effects of a specific variable are difficult to isolate if other simultaneous variables change in the interim due to R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 0 insufficient short-term memory capacity and interference from other variables (Lewis & Anderson, 1985). Finally, delays create instability and oscillations in the system. For example, a decision-maker initiates an action to correct for discrepancies between desired and actual system states. If there is a delay between the initiation of the action and its effects, a decision-maker may fail to account for this delay and continue to initiate corrective actions even after sufficient corrective actions have been taken. As a result of the over-correction, the system oscillates. A person’s ability to control for confounding variables and detect cause and effect relations diminishes with such instability in the system. As such, the time lag between current investments and their impact on the SFM creates difficulties in the learning of relationships between investments and firm profits, which impede performance improvement over time. Different PMSs may also induce different learning modes. Psychological research suggests that there are two different learning modes (Berry & Broadbent, 1988; Hayes & Broadbent, 1988). An individual may use an unselective learning mode where the person observes all variables and the contingencies between them unselectively. Correct as well as incorrect factors are stored. Unselective learning is relatively slow and requires greater experience before the person retains a sufficiently large number of condition-action links that will ensure effective performance. The knowledge acquired also tends to be implicit. Alternatively, an individual may use a selective learning mode where the person focuses only on a few variables and observes the contingencies between these selected variables. The R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 21 knowledge acquired tends to be explicit since only a small number of relationships are involved. If the correct variables are selected, selective learning is fast and effective. However, if incorrect variables are selected, people perform worse under selective learning than under unselective learning. Explicit instructions to search for rules tend to induce selective learning. Otherwise, people are more likely learn unselectively. A SFM alone does not provide an explicit model of cause and effect relations since it is a summary measure. Therefore, managers may be more likely to learn unselectively. Learning may then be slower with managers needing more experience before performance improves. Prior research has indicated that the process by which people assess correlation between variables is non-normative in that people are affected by normatively irrelevant factors such as the organization of variables, semantic referents to variables, and scale transformations of data (Hutchinson & Alba, 1997). If information is presented in a tabular format where every row represents data of multiple variables from a single period and every column represents data of a single variable from multiple periods, people are more likely to make within-row comparisons than within-column comparisons (Hutchinson & Alba, 1997). In this study, within-row comparison (i.e., within the same time period) may result in managers not attending to the relationships between their investments and the SFM within-column (i.e., across time periods) and not picking up the time lags. As such, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 22 they would face greater difficulties learning the underlying relationships between investments and firm profits. Learning from VDs When firms add VDs to a SFM in the PMS, managerial performance may improve because managers learn better and/or managers are following better incentives (i.e., they are better motivated). Learning from having VDs in the PMS may come from three different sources. Learning may arise from VD information, from VD weights, or from paying greater attention to VD information and VD weights once VDs are rewarded. It is possible that the learning effects from VD information and VD weights only manifest themselves when VDs are actually rewarded as managers begin to attend to VD information and VD weights. Learning from VD Information Intermediate causal models that link temporally separated events make it easier for people to infer causality (Einhom & Hogarth, 1986). VDs reflect intermediate value-creating activities that ultimately drive future financial performance. For example, in a firm that relies on innovation for its long-term success, increasing R&D investments will increase R&D productivity before increasing SFM in the future. R&D productivity reflects the value-creating activity of innovation and is the intermediate measure between R&D investments and SFM. Hence, adding VDs to a SFM provides managers with an intermediate causal mental model to test hypotheses about the underlying relationships between investments and firm profits. Intermediate VDs also make the process causing the time delay between Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 investments and financial performance more apparent, thereby encouraging managers to use a feedforward strategy that takes into account the time delay (Brehmer, 1990, 1992; Brehmer & Allard, 1991). Since VDs are leading indicators of financial performance, investments are linked faster to VDs than to the SFM. Therefore, VDs shorten the time lag between the manager investing and the managers receiving feedback about the impact of their investments. As such, VDs increase the likelihood that managers will be able to detect and learn the lagged relations between investments and firm profits as well as the relative importance of different investments. Since adding VDs to a SFM provides a manager with an explicit model of the cause and effect linkages amongst the different performance measures, it may induce the manager to adopt a selective learning mode. If the right VDs are identified in the mental model, adding VDs to a SFM promotes selective learning that is faster and more effective. Also, if people tend to make within-row (i.e., within the same time period) rather than within-column (i.e., across time periods) comparisons in assessing the relationships between investments and firm profits, using VDs that do not lag investments as much as a SFM may ensure that managers attend to the relationships between their investments and VDs (Hutchinson & Alba, 1997). Hence, it may be easier for managers to learn the underlying relationships between investments and firm profits. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 4 Prior research indicates that people have difficulties assessing the impact of an action that is not conceptually related to the outcome when they are making predictions (Broniarczyk & Alba, 1994; Tversky & Kahneman, 1982). Tversky and Kahneman (1982) suggest that people use causal schemas that flow from causes to effects to infer relationships. The prior beliefs of people about the theoretical relationship between an action and an outcome influence the use of that action in predicting the outcome. People tend to view actions that are conceptually related to an outcome in the causal schema to be more predictive of the outcome than actions that are not naturally related to the outcome (Tversky & Kahneman, 1982). A VD may be conceptually more related to firm profits than the monetary amount of an investment. VDs, which are intervening variables between investments and firm profits, make the relationship between investments and firm profits more salient and apparent to managers (Luft & Shields, 2002). People perform better in inference tasks when the cues and criterion used are set in a meaningful environment as compared to an abstract environment (Muchinsky & Dudycha, 1975; Sniezek, 1986). A meaningful task environment makes learning easier. When cues and criterions are given meaningful labels, they suggest to the learner the statistical structure of the relationship. If the suggested statistical structure matches the actual statistical structure, the meaningful labels increase knowledge and improve performance. Meaningful and congruent labels also improve consistency because they reduce the number of strategies that need to be searched and tested. Meaningful labels also improve the encoding and retrieval of data. Hence, adding Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 5 VDs to a SFM may make the task more meaningful and less abstract to managers. Managers may find it more meaningful, for example, to relate the number of new products and operating efficiency to future firm profits than to relate a more abstract R&D investment and plant and equipment (P&E) investment to future firm profits. Luft and Shields (2002) found that people made more accurate predictions of future profits with nonfinancial measures of product quality than with financial measures of product quality even though both measures had similar statistical predictive ability. In a second experiment, subjects allocated resources across two quality programs, one of which had a greater positive impact on future profits. They found that adding nonfinancial measures of product quality to financial measures of product quality resulted in people allocating more resources to the more profitable quality program. Adding nonfinancial measures did not increase the statistical predictive ability but increased the amount of information a subject must process. They attributed the beneficial effect of nonfinancial measures to them being cognitively more meaningful to people than financial measures. Nonfinancial measures activated mental associations that were more relevant for predicting future profits than financial measures. Information Overload with VD Information While the above arguments support the hypothesis that managers learn and perform better when VDs are added to a SFM, the counterargument is that adding VDs to a SFM significantly increases the amount of information that managers must process. The number of cause and effect relations to be processed increases with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 6 VDs being added to a SFM. The cause and effect relations also have different time lags and intercorrelations that complicate information processing. Judgment performance may deteriorate as the number of cues available to a person increases (Lee & Yates, 1992). The chances for error increase as a person has to perform an increasing number of mental operations when the number of cues increases. Moreover, people may shift their attention from one cue to another as more cues are added, decreasing the reliability of their judgments. A person may also become more frustrated with the task as more information is available and give up on the task, leading to deterioration in judgment quality. For example, people may keep making the same decisions regardless of changes in information or they may focus only on a pre-selected set of cues and ignore all other cues. Hence, they may pay less attention to cues that should have received more emphasis. Managers may also be confused by conflicting measures between leading VDs and a lagging SFM. For example, a leading VD may show good performance while a lagging SFM may show poor performance. Managers may then become less sensitive to changes in the VDs and SFM in their judgment of the relationship between investments and firm profits. For example, Peterson and Pitz (1986) asked subjects to predict the number of games major league baseball teams won over a season using cues such as each team’s earned run average, batting average, and home run average for a particular season. They found that the predictions of subjects regressed more towards the mean when the number of cues increased. They attributed this effect to additional cues conflicting with a single cue, causing subjects Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 7 to make less extreme predictions. Lee and Yates (1992) also note that it is possible for variability of judgment to decrease as the number of cues increases. This follows the law of large numbers where the variability of a mean decreases as the sample size increases. On the other hand, a manager’s judgments and decisions may become more variable when VDs are added to a SFM. A person’s judgment is influenced upwards or downwards by different values of various cues and as the number of cues increases, the person’s judgment would shift around more (Lee & Yates, 1992). Hence adding VDs to the SFM may cause managers to change their investments more frequently as they encounter changes in not only the SFM but the VDs as well. Changing judgments frequently may create instability in the system, which impedes the learning of dynamic lagged relations (Sterman, 1994). Therefore, adding VDs to a SFM increases the complexity of the task, further straining the limited cognitive capacities of people. Prior research finds that increased task complexity hinders learning ability and performance (e.g., Bonner, 1994; Shields, 1983; Stocks & Harrell, 1995; Wood, 1986). With a SFM alone, there is a single focus for accountability and confusion that comes from multiple measures is eliminated. In summary, managers may learn better with VDs in the PMS because VD information as compared to SFM provide a better mental model linking investments and firm profits, induces better task strategies, reduces lag time between a managerial action and its effects on performance measures, promotes better learning, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 8 and are cognitively more meaningful. However, adding VDs to a SFM may potentially decrease learning since it significantly increases the amount of information to be processed and the complexity of the task. Learning from VD Weights Managers may also learn better with VDs in the PMS if VDs are weighted accurately in the incentives system to reflect the relative importance of different investments. Prior multiple cue probability learning (MCPL) research has shown that feedback about the nature of the relationship between the cues and the criterion improves judgment performance (e.g., Balzer, Doherty, & O’Connor, 1989; Balzer, Sulsky, Hammer, & Sumner, 1992; Remus, O’ Connor, & Griggs, 1996; Summers & Hammond, 1966; Todd & Hammond, 1965). For example, Todd and Hammond (1965) found that feedback on the correct weights of the cues and the actual cue weights used by the subject improved the performance of the subject relative to just providing information about the correct value of the outcome variable. There are various categories of feedback. Outcome feedback refers to information about the actual outcome of the criterion. Cognitive feedback, on the other hand, refers to information about interrelationships between cues, criterion, and judgments. Cognitive feedback can be further classified into three categories: task information, cognitive information, and functional validity information (Balzer et al., 1989). Task information refers to information about the task environment, that is, relationships between the cues and the criterion. Cognitive information refers to information about the person’s judgment strategy, that is, relationships between the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 9 cues and the person’s judgment. Functional validity information refers to information about the relationships between the task environment and the person’s cognitive system, that is, between the person’s judgments and the criterion. In a review of empirical research on the effects of cognitive feedback on judgment performance, Balzer et al. (1989) conclude that task information is the component of cognitive feedback that improves judgment performance. In a more recent study, Remus et al. (1996) extended the findings of Balzer et al. (1992) to a judgment task that involved structural instabilities. Remus et al. found that task information feedback improved the accuracy of recurring forecasts of time series data as compared to outcome feedback. In Remus et al., subjects were first shown historical time series data with 20 periods that was basically flat with some error term. Subjects then forecasted time-series data with eight periods that was a continuation of the first time series. Finally, they forecasted time-series data that increased at 2% per period, decreased by 2% per period, or remained constant for 10 periods. Subjects were assigned to different conditions with different types of feedback: simple outcome feedback; graphical performance outcome feedback, qualitative performance outcome feedback, task information feedback, and task plus cognitive information feedback. Luft and Shields (2002) found that replacing raw data of a financial measure of product quality with a statistical model of how the financial measure was related to future profits improved the accuracy of future profits predictions. They attributed this effect to the statistical model removing the subjective processing of raw data, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 0 thereby improving judgment accuracy. In the second experiment, they found that including a statistical model of how financial and nonfinancial measures of product quality were related to future profits with the raw data also increased the allocation of resources to the more profitable quality program. Hence, managers may improve their learning of the relative importance of investments through VD weights that are provided. Moderating Effect ofFirm Type on Learning from VDs The type of firm may moderate the effects on managerial learning of adding VDs to a SFM. I study two types of firms: firms that rely more on tangible assets for their competitive advantage, tangible assets firms (TAFs); and firms in which intangible assets firms play a more significant role in their long-term success, intangible assets firms (ITAFs). The learning effects from having VD information and VD weights may be more significant in an ITAF than in a TAF. This is because a SFM more adequately reflects investments in tangible assets than investments in intangible assets. Under GAAP, investments in tangible assets are typically capitalized whereas investments in intangible assets are immediately expensed. In the short term, a SFM will be more negatively affected by investments in intangible assets than by investments in tangible assets because investments will not pay off until sometime in the future. As such, giving SFM feedback alone may lead managers to perceive that investments in tangible assets are more beneficial for firm profits than investments in intangible assets. Such a perception conforms to reality for a TAF but is contrary to reality for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 an ITAF. Hence, in an ITAF, VD information and VD weights that weigh intangible assets more heavily may correct the wrong perception portrayed by a SFM alone. Managers of an ITAF are expected to learn better that intangible assets are actually more important than tangible assets with VD information than without VD information and with VD weights than without VD weights. For managers of a TAF, a SFM alone already inclines them to accurately perceive that tangible assets are more important than intangible assets. In a TAF, VD information and VD weights that weigh tangible assets more heavily will not cause managers to learn much more than what they already know with only a SFM. Heuristics Used in Learning Hutchinson and Alba (1997) propose that people use different types of heuristics to learn correlations among variables and that the particular heuristic used is dependent on the context. They considered three major categories of heuristics: exemplar-based, chunk-based, and difference-based heuristics. Exemplar-based representation assumes that people reason by analogy to specific instances. Chunk- based representation assumes that people group observations to simplify information processing. Difference-based representation assumes that people compare specific changes in the outcome variable to the corresponding changes in the decision variables. For each major category of heuristics, people can sample observations based on either an informational criteria or a locational criteria. An informational criteria selects observations based on their values whereas a locational criteria Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 2 chooses observations based on their spatial characteristics. In total, Hutchinson and Alba identified 12 heuristics (Table 1). Table 1 Classification of Heuristics Heuristic representation Sampling Criteria Informational Locational Exemplar-based Best exemplar heuristics 1) Best exemplars 2) All exemplars Primacy-recency heuristics 3) Recent heuristic 4) Initial heuristic Chunk-based Prototype heuristics 5) Best vs worst prototypes 6) Many vs few chunks Trend-detection heuristics 7) Recent vs initial prototypes 8) Many vs few chunks Difference-based Size-based heuristics 9) Large outcome differences 10) Small outcome differences Proximity-based heuristics 11) Adjacent rows differences 12) Within groups differences The investigation of the use of different types of heuristics is exploratory in nature. I have no predictions as to the differences in usage of heuristics across different PMSs and different firm types. Effects of a PMS on Motivation In addition to its effects on learning, a PMS also affects the motivation of managers since it determines the performance measures in the incentive contracts with managers. The informativeness principle discussed by Holmstrom (1979) suggests that a performance measure will improve an incentive contract if it provides information about a manager’s actions above that provided by the other performance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 measures in the contract. Two important properties of a performance measure are its congruence with the principal’s desired payoff and its noise that is unrelated to the agent’s actions (Banker & Datar, 1989). Theoretical work in multi-task agency settings shows that an additional performance measure in an incentive contract is valuable if it reduces noncongruity or noise (e.g., Feltham & Xie, 1994). Hence, if VDs provide additional information relative to a SFM with respect to desirable managerial actions in executing firm strategies, then adding VDs to a SFM will improve the incentive contract. Incentives may be better with VDs being rewarded in the PMS if the SFM alone is an insufficient statistic to motivate managers to allocate more resources to the relatively more important investment. It has been argued that providing rewards contingent on performance reduces the intrinsic motivation for a task and performance because such rewards reduce people’s perception of competence and self-determination (Deci & Ryan, 1985). However, in a meta-analysis of 96 studies, Eisenberger and Cameron (1996) found that the negative effects of rewards on intrinsic motivation existed only under very limited conditions in the laboratory that were atypical in real life. In fact, people who were initially promised rewards for achieving a performance standard increased the amount of time spent on a task after the rewards were subsequently removed and had more favorable attitudes towards their task (Eisenberger & Cameron, 1996, 1998). Eisenberger, Rhoades, and Cameron (1999) also found evidence supporting the hypothesis that performance-contingent rewards improved the perception of autonomy and intrinsic motivation. Hence, in the context of my study where Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. managers are rewarded based on the level of their performance, I do not expect their intrinsic motivation to be reduced. Motivation from Rewarding SFM Rewarding only a SFM may cause managers to pay too little attention to the tasks that are not adequately captured by the SFM. In a multi-task agency setting, Holmstrom and Milgrom (1991) show that an incentive contract that rewards performance based on a measure that does not adequately capture performance on all desirable tasks may lead to suboptimal effort on those tasks in which performance are more difficult to measure. When firm value is created primarily through tangible assets such as plant, equipment, property, and inventory, traditional financial measures adequately capture investments in these assets and the expenses associated with the usage of these assets to generate revenues. However, when firm value is created primarily through intangible assets such as innovation in products, services, and processes; human resources; and customer relationships, traditional financial measures do not adequately measure investments in these assets. For example, spending on intangibles such as R&D and human resources development is expensed immediately under GAAP. Financial measures do not reflect the value drivers and the strategies of firms that rely on intangibles. When a SFM alone is rewarded, managers will be motivated to allocate more resources to develop tangible assets relative to intangible assets regardless of the type of firm they are managing. Since investments in intangible assets are typically expensed under GAAP while investments in tangible assets are capitalized and permission of the copyright owner. Further reproduction prohibited without permission. 35 expensed over multiple periods, the former will reduce a SFM and the monetary rewards based on the SFM in the short term more than the latter. Hence, when only SFM is rewarded, managers of TAF are expected to be better motivated to direct more resources to the more important investment than managers of ITAF. Motivation from Rewarding VDs When VDs are added to a SFM and rewarded in the PMS, managers are motivated to allocate their resources to increase those VDs according to their respective weights in the PMS so as to increase their monetary rewards. In addition to the motivational effects of rewarding VDs, it is possible that merely providing feedback on VDs and assigning goals on the VDs may motivate managers to work towards improving their performance on those VDs. Locke and Latham (1990, chap. 8) review studies of the effects of goals and feedback on performance and conclude that feedback motivates people through the process of goal setting and that goals motivate people more effectively when feedback is present. In studies where feedback is provided for multiple dimensions of performance, people tend to do better on those dimensions for which goals have been set (e.g., Kolb & Boyatzis, 1970; Nemeroff & Cosentino, 1979; Schmidt, Kleinbeck, & Brockmann, 1984). Other studies showed that when goals are set for tasks, performance tend to improve in the presence of feedback but not so in the absence of feedback (e.g., Becker, 1978; Erez, 1977; Komaki, Barwick, & Scott, 1978; Strang, Lawrence, & Fowler, 1978). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 6 Moderating Effect o f Firm Type on Motivation from Rewarding VDs In a TAP, VDs with accurate weights would emphasize tangible assets more than intangible assets. However, rewarding a SFM alone already inclines managers to allocate more resources on developing tangible assets relative to intangible assets. Therefore, rewarding accurately-weighted VDs to a SFM will not motivate managers to allocate more resources on developing tangible assets than without rewarding VDs. A SFM is a sufficient statistic to motivate managers to allocate more resources on developing tangible assets than intangible assets. Accurately-weighted VDs are not more congruent with firm performance and do not provide significant additional information about desirable managerial actions relative to a SFM alone. In a TAF, managerial motivation and performance are not expected to improve with the rewarding of accurately-weighted VDs than without the rewarding of VDs. In an ITAF, VDs with accurate weights would emphasize intangible assets more than tangible assets. Therefore, managers will be motivated to allocate more resources on developing intangible assets with the rewarding of accurately-weighted VDs than without the rewarding of VDs. A SFM is not a sufficient statistic to motivate managers to allocate more resources on developing intangible assets relative to tangible assets. Accurately-weighted VDs are more congruent with firm performance and provide significant additional information about desirable managerial actions relative to a SFM alone. Therefore, in an ITAF, managerial motivation and performance are expected to improve with rewarding accurately- weighted VDs than without rewarding VDs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 7 Hypotheses In this study, I try to disentangle the learning effect from having VD information, the learning effect from having accurate weights on VDs, the learning effect from rewarding VDs, and the motivation effect from rewarding VDs, when VDs are added to a SFM in the PMS. The differences in learning and motivation across conditions are hypothesized to lead to differences in resource allocation decisions and performance across conditions. While I choose to specify the direction of my hypotheses for clarity, this study is exploratory in nature and the actual pattern of results will be more informative than the hypothesized pattern of results. The individual hypotheses are discussed below. The first set of hypotheses, HI to H4, focuses on the effects of the independent variables (PMS and firm type) on the ultimate dependent variable, managerial performance. The hypothesized results for HI to H4 are depicted in Figure 3. HI: When rewarded with SFM alone, TAF managers perform better than ITAF managers. H2 (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their performance more than TAF managers do. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 8 H3 (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their performance more than TAF managers do. H4 (Effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their performance more than TAF managers do. The performance differences in H4 could be a result of motivation effect or learning effect from rewarding VDs. This study is designed to disentangle these two effects. Learning is measured separately in a post-experimental questionnaire and is used as a covariate in an ANCOVA on performance to control the effects of learning and isolate the effects of motivation. TAF A Learning from VD w weinhts A Learning from VD information Effect from VD rewards Reward Reward SFM Reward SFM Reward SFM SFM alone alone and provide alone and provide and VDs VD information VD information and VD weights Figure 3. Graphical depiction of HI to H4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 9 The second set of hypotheses, H5 to H8, focuses on the effects of the independent variables (PMS and firm type) on the mediating dependent variable, learning. H5: When rewarded with SFM alone, TAF managers learn better about the relationships between investments and firm profits than ITAF managers. If H5 is not supported in that TAF managers do not learn better about the relationships between investments and firm profits than ITAF managers, then it would suggest that motivation rather than learning is driving the results in HI. If H5 is supported, then any performance differences in HI may be a result of either learning differences or motivation differences. H6a (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. In H6a, better learning when VD information is provided may be a result of increased attention of managers on the effects of the more important investment or the long-term impact of their investments on performance, which are tested in H6b and H6c, respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 H6b (Attention to the more important investment from having VD information): When VD information is provided as compared to when it is not, managers pay more attention to the effects of the more important investment on performance. H6c (Attention to the long-term impact of investments from having VD information): When VD information is provided as compared to when it is not, managers pay more attention to the long-term impact of investments on performance. H7a (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. In H7a, better learning when VD weights are provided may be a result of increased attention of managers to VD information, to the effects of the more important investment on performance, or to the long-term impact of investments on performance, which are tested in H7b, H7c, and H7d, respectively. H7b (Attention to VD information from having VD weights): When VD weights are provided as compared to when they are not, managers pay more attention to VD information. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 H7c (Attention to the more important investment from having VD weights): When VD weights are provided as compared to when they are not, managers pay more attention to the effects of the more important investment on performance. H7d (Attention to the long-term impact of investments from having VD weights): When VD weights are provided as compared to when they are not, managers pay more attention to the long-term impact of investments on performance. H8a (Learning effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. If all the learning effects are from having VD information and VD weights (i.e., H6a and H7a are supported), H8a will not be supported. However, it is possible that when VD information and VD weights are provided but VDs are not rewarded, managers do not attend to VD information, VD weights, and the long-term impact of their investments on performance. As such, managers do not learn from having VD information and VD weights until the VDs are actually compensated. If that is the case, H2 (and the corresponding H6a) and H3 (and the corresponding H7a) will not be supported whilst H4 and H8a will be supported. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 In H8b, H8c, H8d, I test whether rewarding VDs leads to increased attention to VD information, the effects of the more important investment on performance, and the long-term impact of investments on performance. H8b (Attention to VD information from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to VDs information. H8c (Attention to the more important investment from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to the effects of the more important investment on performance. H8d (Attention to the long-term impact of investments from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to the long-term impact of investments on performance. The third set of hypotheses, H9 to H12, focuses on the effects of independent variables (PMS and firm type) on the mediating dependent variable, the optimality of resource allocation decisions. H9: When rewarded with SFM alone, TAF managers allocate resources better than ITAF managers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 3 H10 (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their allocation of resources more than TAF managers do. Fill (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their allocation of resources more than TAF managers do. H12 (Effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their allocation of resources more than TAF managers do. In summary, the hypotheses test whether TAF managers learn, allocate resources, and perform better than ITAF managers under SFM alone. The hypotheses also test whether VD information, VD weights, and VD rewards improve the learning, resource allocation, and performance of ITAF managers more than TAF managers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 4 CHAPTER 3: EXPERIMENTAL DESIGN AND PROCEDURES Experimental Design The experiment is a 2 X 4 factorial design (Table 2). The first between- subjects factor is the type of Firm: TAF and ITAF. The second between-subjects factor is the type of PMS: Reward SFM alone (PMS1), Reward SFM alone and provide VD information (PMS2), Reward SFM alone and provide VD information and VD weights (PMS3), and Reward SFM and accurately-weighted VDs (PMS4). The experiment is designed to separate the learning effect from having VD information, the learning effect from having VD weights, the learning effect from rewarding VDs, and the motivation effect from rewarding VDs. Learning and motivation effects are inferred primarily from performance differences across conditions. Performance differences due to differences in feedback suggest learning differences while performance differences due to differences in incentives suggest motivation differences. In addition, post-experimental questions are asked to measure the knowledge of subjects about the relationships between investments and firm profits. The difference in the knowledge of subjects across conditions fiirther disentangles the learning and motivation effects. Comparing PMS2 and PMS1 isolates the learning effect of just having VD information since there is no difference between PMS2 and PMS1 other than PMS2 providing information on the performance of VDs. The feedback provided differs between PMS2 and PMS1 but the incentive systems are identical. Hence, any Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 performance differences between PMS2 and PMS1 are due to the differences in feedback and not differences in incentives. Comparing PMS3 and PMS2 isolates the learning effect of just having VD weights since the only difference between PMS3 and PMS2 is that PMS3 informs subjects of the appropriate relative weights of the two types of investments. The incentive systems are similar in PMS3 and PMS2. Any performance differences between PMS3 and PMS2 are due to differences in feedback rather than incentives. Finally, comparing PMS4 and PMS3 isolates the effect of rewarding VDs since PMS4 and PMS3 differs only in that PMS4 actually rewards subjects according to their performance on VDs. In PMS4, the performance of the VDs is weighted in the incentives system to reflect the relative importance of the two types of investments. Hence, the incentive systems are different in PMS4 and PMS3 but the information provided is identical. Therefore, any performance differences between PMS4 and PMS3 are due to differences in incentives rather than feedback. Type and Number of Subjects 163 MBA students from the University of Southern California participated in the experiment. Subjects were recruited by going into their classrooms to advertise the experiment as well as via email messages. The average age of the participants was 30.80 (SD = 5.37) with an average of 7.72 (SD = 5.11) years of working experience. 80.25% of the subjects were male and 19.75% of the subjects were female. Of the 163 subjects, 157 subjects had taken at least one financial accounting course (M = 1.26, SD = 0.68) and 137 subjects had taken at least one management Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 6 accounting course (M = 0.99, SD = 0.61). Table 2 presents the descriptive statistics on the demographic variables of subjects in each condition of the experiment. Table 2 Experimental Design and Descriptive Statistics Firm Reward SFM alone (PMS1) Reward SFM alone and provide VD information (PMS2) Reward SFM alone and provide VD information and VD weights (PMS3) Reward SFM and accurately- weighted VDs (PMS4) TAF Condition 1 Condition 2 Condition 3 Condition 4 n = 19 « = 21 n - 20 n = 21 age = 30.32 age = 28.90 age = 29.20 age = 32.52 wyears = 7.32 wyears = 5.86 wyears = 6.70 wyears = 9.19 facourse = 1.26 facourse = 1.53 facourse =1.10 facourse = 1.48 macourse = 0.94 macourse = 1.25 macourse = 0.90 macourse = 1.05 17 males 18 males 12 males 17 males 2 females 3 females 8 females 4 females ITAF Condition 5 Condition 6 Condition 7 Condition 8 n = 21 n= 19 w = 21 « = 21 age = 31.43 age = 30.17 age = 31.76 age = 31.90 wyears = 8.48 wyears = 7.94 wyears = 7.90 wyears = 8.71 facourse = 1.24 facourse = 1.11 facourse = 1.19 facourse =1.14 macourse = 0.81 macourse = 0.94 macourse =1.00 macourse = 1.00 15 males 17 males 16 males 18 males 6 females 1 female 1 missing 5 females 3 females Note, n = Number of subjects in each cell; age = Mean age; wyears = Mean years of working experience; facourse = Mean number of financial accounting courses taken for MBA degree; macourse = Mean number of management accounting courses taken for MBA degree. The sex (F ~ 1.56, p = .15), age (F = 1.28, p = .26), years of working experience (F = 0.97, p = .45), number of financial accounting courses taken for MBA degree (F = 1.22, p = .29), and number of management accounting courses taken for MBA degree (F — 0.93, p = .49) of subjects were not significantly different Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 7 across conditions. Differences in the dependent variables across conditions were not likely to be a result of demographic differences of subjects. Experimental Procedures The experimental instrument was delivered via a web-based program (see Appendix A for the experimental stimuli). Recruited subjects logged on to the website with a unique log-in name and password. Subjects were informed that they were participating in a study where they would assume the role of a division manager of a company managing four divisions. They would manage each division for an unknown number of periods before being transferred to the next division.6 They were told that each division was similar to but independent from the other divisions and that they could use their experience in one division to help them manage the other divisions. Subjects were then given background information about the company and task instructions. Each period, subjects had a fixed budget of $100 million that they had to spend entirely on R&D and/or P&E. Subjects were told that their objective was to maximize the operating income before deduction of R&D expenses and depreciation (hereafter referred to as “gross operating income”).7 Unknown to the 6 The total number of periods in each division was not disclosed to the subjects before they began managing the division to prevent the subjects from attempting to manipulate the SFM towards the last few periods o f a division. Subjects could increase the SFM significantly by reducing the R&D investment to zero during the last few periods o f a division since R&D investment was immediately expensed in the period invested and the adverse effects of reducing R&D investment to zero would be delayed until after they had finished managing the division. 7 Gross operating income was used as the objective for firm performance because it captured the economic effects of the investments on firm profits in the experiment and disregarded the effects of different accounting treatments o f R&D investment (immediate expensing) and P&E investment Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 8 subjects, R&D and P&E investments were differentially related to gross operating income with a lag of three periods.8 A lag would exist between R&D investment and its impact on gross operating income since time would be required to bring basic research to commercialization. A lag between P&E investment and its impact on gross operating income would occur because time could elapse between the investment and the installation of new capital equipment and between the installation and the optimal usage of new capital equipment. Next, subjects answered some questions about their prior beliefs as to the relative impact of R&D and P&E investments on gross operating income in the current period and future periods. To clear the short-term memory of the subjects, demographic questions were then asked. Subjects were then randomly assigned to one of the eight experimental conditions. Depending on the condition that the subjects were assigned to, the subjects were presented different information on the type of feedback that they would receive each period and their compensation scheme. Subjects were also informed of the R&D and P&E investments in the prior four periods that were invested by their (capitalizing and depreciating). In other words, gross operating income represents firm profits in this study. 8 Different lags relating investments to firm profits can occur in the natural environment. A lag of three periods is plausible. Mean R&D time lags of 3.7 years for Chemicals, 3.0 years for Motor Vehicles and Transportation Equipment, 2.6 years for Industrial Machinery, and 2.1 years for Electrical Machinery, have been documented in prior studies (Suzuki, 1985). Luft and Shields (2001) used a lag of three periods between quality expenditures and profits. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 9 predecessors since gross operating income and SFM in the first three periods of a division were affected by these prior R&D and P&E investments.9 Before subjects began managing the four divisions, they practiced managing a division for five periods. The practice division was similar to the actual four divisions except that subjects did not receive any compensation for their performance in the practice division. Subjects first decided how much to spend on R&D and P&E in each period. After the decision, subjects were shown their R&D and P&E investments, actual gross operating income, SFM, and points awarded for that period. In the PMS2, PMS3, and PMS4 conditions, subjects were also shown their performance on the two VDs that reflected performance on R&D productivity and P&E operating efficiency. Subjects could also view the same information for all prior periods of that division and all prior divisions that they had finished managing. Subjects then made their decisions for the next period. There were 17 periods in the first division, 15 periods in the second division, 15 periods in the third division, and 14 periods in the fourth division. After the final period of each division, subjects were shown the total points they accumulated for that division. Prior accounting studies that examined judgment performance in assessing lagged relationships used passive forms of presenting information to subjects (Luft & Shields, 2001, 2002). Typically, subjects would be presented learning data consisting 9 Gross operating income in the first three periods was determined by R&D and P&E investments in the prior three periods due to the three-period lag between investments and gross operating income. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 of expenditures and profits for multiple time periods by the experimenter before predicting profits for a new set of expenditures. Subjects in Experiment 1 of Luft and Shields (2002) and in Luft and Shields (2001) learned from quality and profits data for four time periods (quarter -3 to quarter 1) for 20 plants similar to their own 20 plants passively before making their profit predictions for their own 20 plants with quality data for four time periods. Subjects in Experiment 2 of Luft and Shields (2002) also learned from data for four time periods (4 years) for 20 plants similar to their own plant passively before making resource allocation decisions for their own plant. Prior research indicates that subjects with active control over cue combinations and resultant feedback improve their judgment performance compared to subjects with passive control where cue combinations and feedback are selected by the experimenter (Lindberg & Brehmer, 1977a, 1977b). Active feedback control encourages subjects to actively test their hypotheses about the relationships between cues and outcomes, which improves learning and performance (Lewis & Anderson, 1985). Hence, in my experiment, subjects actively made investment decisions and learned from the data sequentially for 14 to 17 time periods for the four divisions that they were managing. Moreover, this procedure generalizes the results of my study to situations where managers learn from actively making decisions and observing the impact of their decisions on their own division sequentially over time. The results of prior studies such as Luft and Shields (2001) and Luft and Shields (2002) only generalized SFM (ROI) in the first period was also affected by P&E investment in the prior four periods because Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 to situations where managers learned from observing data simultaneously from multiple divisions similar to their own. Also, the number of time periods provided to subjects in these earlier studies might strongly suggest to the subjects that the maximum lag was three time periods. In my experiment, the number of time periods would not suggest to the subjects the number of periods in the time lag. After subjects finished managing the fourth division, they answered a post- experimental questionnaire containing manipulation check questions; questions on their relative attention to performance measures, the effects of R&D versus P&E investments on performance, and the short-term versus long-term effects of investments on performance; questions on their knowledge about the relationships between investments and gross operating income; and questions on their strategies used to understand the relationship between investments and performance. There were two versions of the post-experimental questionnaire with identical questions but different ordering of questions as well as different ordering of options in questions with multiple options. This procedure mitigated potential order effects. Appendix A shows the first version of the post-experimental questionnaire for Condition 1. Operationalizations of Independent Variables SFM The SFM was operationalized as the return on the total amount of investment each period (ROIt). P&E investment was capitalized and depreciated over five periods. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ROIt = (7 rt - Depreciationt - R&Dt)/(Invested Amount). 52 (1) 7 it = Gross operating income, i.e., operating income before depreciation and R&D expenses in period t (in $ million). Invested Amount = R&Dt + P&Et = $100 million. R&Dt = R&D investment in period t (in $ million), 0 < R&Dt < 100. R&Dt is expensed in period t. P&Et = P&E investment in period t (in $ million), 0 < P&Et < 100. P&Et is capitalized and depreciated over 5 periods. Depreciationt = P&Et.4 /5 + P&Et-3/5 + P&Et_ 2 /5 + P&Et_i/5 + P&Et/5. Subjects were given information on how ROIt is calculated. To avoid subjects inferring from the difference in the accounting treatments of R&D and P&E investments about the relationships between investments and gross operating income, subjects were told that the company’s “choice of whether to expense or capitalize R&D and P&E investments is in accordance with generally accepted accounting principles and may not reflect actual economic reality.” VDs There were two VDs in the experiment, R&D productivity and P&E operating efficiency. R&D Productivity The VD for intangible assets, R&D productivity, was operationalized as the number of new product introductions. Number of new product introductions was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 positively correlated with R&D investment and there was no time lag between R&D investment and the number of new product introductions in the experiment. In reality, there will be a time lag between R&D investment and the number of new product introductions. This time lag will most likely be shorter than the time lag between R&D investment and gross operating income since VDs are leading indicators of financial performance. R&DPt = Number of new product introductions in period t (rounded to the nearest figure), 0 < R&DPt < 46. P&E Operating Efficiency The VD for tangible assets, P&E operating efficiency, was operationalized as the operating efficiency rating given by an external consultant engaged by the firm to audit its manufacturing facilities. Operating efficiency rating was positively correlated with P&E investment and there was no time lag between P&E investment and operating efficiency rating in the experiment. In reality, there will be a time lag between P&E investment and operating efficiency, which will most probably be shorter than the time lag between P&E investment and gross operating income. R&DPt = 10 ln(R&Dt+ 1). (2) P&EOEt = 20 ln(P&Et + 1). (3) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 4 P&EOEt = Operating efficiency rating in period t (rounded to the nearest figure), 0 < P&EOEt< 92. A natural log function is used to relate R&Dt to R&DPt and P&Et to P&EOEt because it ensures that R&Dt and P&Et have positive and diminishing marginal rates of return to R&DPt and P&EOEt, respectively, reflecting reality. Appendix B describes in detail the properties of the natural log function that makes it suitable for my study. Firm (First Factor) Description o f Industry and Firm in Experiment The experimental company was a mid-size firm in the semiconductor industry (SIC 3674). The semiconductor industry is chosen because both R&D and P&E investments are vital to firm performance (Smith, 2002; Tortoriello, 2002). The Semiconductor Industry Association reported that the US semiconductor industry spent $13.3 billion (18.5% of sales) on R&D and $13 billion (18% of sales) on capital equipment in 2001 (Semiconductor Industry Association, n.d.). Hence, the semiconductor industry will not overtly suggest that either R&D or P&E investment is relatively more important for firm performance. Technological advancements is critical in the industry as chip makers strive to improve the performance of chips through reducing the size of chips, improving the reliability of interconnects, and increasing the size of silicon wafers. As a result of the fast pace of technological advancement, capital equipment has an average useful life of 3 to 5 years in the industry (Sieling, 1988). It is important for capital Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 equipment to be constantly replaced for the production of new generations of chips. This fits with the assumption in the experiment of the need for constant P&E investments and the use of five periods for the depreciation of P&E investments. Firms in the semiconductor industry also have different strategies with respect to their emphasis on R&D or P&E (Smith, 2002). For example, some semiconductor firms outsource manufacturing operations to devote more resources on R&D (i.e., fabless firms). Other firms choose to build their own wafer fabrication plants to be able to meet strong product demand. Compustat data of semiconductor companies in the S&P MidCap 400 Index from 1991 to 2000 indicates that there are firms that invest more on R&D relative to P&E as well as firms that invest more on P&E relative to R&D.1 0 This agrees with the experimental design, which has different types of firms with different degrees of reliance on R&D or P&E. The background of the experimental company was written based on a survey of the annual reports and company web sites of semiconductor firms included in the S&P MidCap 400 Index. I extracted financial data of the S&P MidCap 400 semiconductor firms from Compustat from 1991 to 2000. The range of possible values of gross operating income, R&D investment, and P&E investment in the experiment existed in the range of actual values of these variables in these mid-size firms. 1 0 The S&P MidCap 400 Index is used by US managers and pension-plan sponsors to measure the performance o f the mid-size company segment o f the US market. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 6 Relationship between Investments and Gross Operating Income A natural log function related the R&D and P&E investments to gross operating income with a lag of three periods (see Appendix B for the properties of the natural log function that makes it suitable for my study).1 1 7 tt = a ln(R&Dt - 3 + 1) + b ln(P&Et - 3 + 1). (4) 7tt = Gross operating income in period t (in $ million). R&Dt - 3 = R&D investment in period t-3 (in $ million). P&Et.s - P&E investment in period t-3 (in $ million). TAF (first level). The competitive advantage of a TAF lies in its tangible assets. A TAF was operationalized as one whose future financial performance was affected more by P&E investment than R&D investment, a and b were arbitrarily set such that (a) b > a, with a = 8 and b = 26 in Equation 4, and (b) reasonable returns were generated from investments in R&D and P&E. 1 1 The use of a nonlinear log function to relate R&D and P&E investments to gross operating income may create learning difficulties for subjects. MCPL literature indicates that the learning of nonlinear cue-criterion relations is slower and less effective than the learning of linear relations (e.g., Brehmer, 1974; Deane, Hammond, & Summers, 1972; Hammond & Summers, 1965; Summers, Summers, & Karkau, 1969). Linear functions are more available than nonlinear functions and hence are accessed and tested earlier and more frequently (Brehmer, 1974). In addition, subjects have difficulties applying their acquired knowledge about the task properties consistently in a nonlinear task (Deane et al., 1972). However, despite the learning difficulties, subjects can improve gradually especially with appropriate task instructions (e.g., Deane et. al, 1972; Hammond & Summers, 1965; Hammond, Summers, & Deane, 1973; Summers & Hammond, 1966; Summers et al., 1969). Appropriate task instructions typically inform subjects about the existence of nonlinear cues or more specifically, the nonlinear function relating the cues and the criterion. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 7 ITAF (second level). The competitive advantage of an ITAF lies in its intangible assets. An ITAF was operationalized as a one whose future financial performance was affected more by R&D investment than P&E investment, a and b were arbitrarily set such that (a) a > b , with a = 26 and b = 8 in Equation 4, and (b) the relative weights for R&Dt - 3 and P&Et - 3 were opposite those for a TAF. Table 3 shows examples of combinations of R&Dt - 3 and P&Et_ 3 , and resultant nt, R&DPt - 3 and P&EOEt_ 3 . Table 3 Sample Combinations of R&Dt - 3 and P&Et_ 3 ; and resultant nt, R&DPt - 3 and P&EOEt - 3 R & D ,_3 ($m) P&Et.3 ($m) R&DPt_ 3 P&EOEt_ 3 7 * for TAF ($m) 7 tt for ITAF ($m) 0 100 0 92 119.99 36.92 10 90 24 90 136.47 98.43 20 80 30 88 138.61 114.31 23 77 32 87 138.70 117.48 30 70 34 85 138.30 123.39 40 60 37 82 136.59 129.44 50 50 39 79 133.68 133.68 60 40 41 74 129.44 136.59 70 30 43 69 123.39 138.30 77 23 44 64 117.48 138.70 80 20 44 61 114.31 138.61 90 10 45 48 98.43 136.47 100 0 46 0 36.92 119.99 Note. For a TAF (ITAF), optimal R&Dt_ 3 = $23 million ($77 million), optimal P&Et_ 3 = $77 million ($23 million) and optimal nt = $138.70 million ($138.70 million). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 8 Appendix C shows the calculations of optimal R&Dt- 3 , P&Et- 3 , and ut for both a TAF and an ITAF. Appendix D shows the calculations of expected values of T c t, R&Dt, P&Et, R&DPt, P&EOEt, assuming a uniform probability distribution for R&Dt and P&Et. PMS (Second Factor) General Description o f PMS In each period, the compensation scheme rewarded and deducted points for every percent of achievement above and below 80% of planned performance on the rewarded measures, respectively. No points were awarded for achieving 80% of planned performance on the rewarded measures. Planned ROIt was the optimal ROIt of 39%.1 2 Planned R&DPt was the maximum R&DPt of 46 products and planned P&EOEt was the maximum P&EOEt of a 92 rating.1 3 In this study, when feedback on ROIt, R&DPt, or P&EOEt was provided, the planned goal for that measure was concurrently provided, regardless of whether the measure was rewarded. This procedure strengthens the test of the performance effects of feedback since prior literature has shown that the process of goal setting mediates the effects of feedback and that feedback moderates the effects of goal setting (Locke & Latham, 1990). The setting of performance targets and the rewarding of performance above the targets are common features in incentive schemes (Merchant, 1989, p. 12; Simon, 1 2 Optimal ROIt was not the maximum ROIt achievable in a period but it was the ROIt if the optimal t c , was achieved across all the periods of a division, rounded to the nearest figure. 1 3 Maximum R&DPt and P&EOEt were not the optimal R&DP, and P&EOEt. Optimal R&DPt and P&EOEt were the R&DPt and P&EOE, that generated the optimal nt. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 9 2000, p. 245-246). Performance targets communicate the desired level of achievement to employees and motivate employees when the targets are tied to bonuses (Simons, 2000, p. 231-233). Prior research suggests that targets should be set at a challenging but achievable level for better performance (Locke & Latham, 1990, chap. 2; Merchant, 1989, p. 26-27; Simons, 2000, p. 242). Meta-analyses of the relationship between goal difficulty and performance showed mean effect sizes ranging from .55 to .82, (Mento, Steel, & Karren, 1987; Tubbs, 1986; Wood, Mento, & Locke, 1987). The number of studies in these meta-analyses ranged from 56 to 72 and the number of subjects ranged from 4,732 to 7,548. A challenging target provides employees with sufficient pressure to perform well and results in more effort and perseverance. An achievable target prevents employees from giving up prematurely. As such, subjects in the experiment were informed that “historical performance in each division indicates that 80% of planned [performance] is an achievable target.” At the end of the study, the total points awarded to the subject over the four divisions were converted proportionately to cash.1 4 The average cash payoff in the study was $13.89, including $5 for completing the study. Subjects were only told of the points rewarded throughout the experiment and not the converted cash award. This avoided the subjects getting disappointed with too small an amount of cash award and helped them to focus on maximizing the number of points. The expected 1 4 To calculate the cash reward, the number of points earned was divided by 200,000 for TAF subjects and by 75,000 for ITAF subjects. This made the actual monetary compensation more equivalent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 0 total points earned under all conditions, assuming uniform probability distribution of R&Dt and P&Et, were approximately the same.1 5 This ensured that performance differences across conditions were not a result of differences in expected points earned but were a result of the manipulations in the experiment. Appendix E shows the calculations of expected points reward for each condition. Incorporation o f Negative Consequences in PMS In practice, penalty provisions in incentive schemes are rare but do exist (Anonymous, 1981; Merchant & Wu, 1995). Practitioner literature has for a long time criticized the lack of downside risk in the incentive contracts of mangers (Cook, 1987; Carol, 1986). Cook (1987) commented that “we have reached a stage where the only penalty in executive compensation is the absence of reward. This is not enough to motivate behavior” (p. 28). Grinblatt and Titman (1987) argued that performance contracts that lack penalties for negative performance encouraged gaming behavior in fund managers. Waller and Bishop (1990) compared a bonus scheme with a bonus-cum-penalty scheme. The bonus scheme rewarded managers linearly based on actual profits of the unit managed. The bonus-cum-penalty scheme is similar to the bonus scheme but penalized managers for unfavorable profit variances. Managers conveyed information about the expected profit function of the unit to a central management that would subsequently allocate resources to the manager for investment. Waller and Bishop (1990) found that there was greater across conditions. Pilot studies indicated that ITAF subjects tended to earn much less points than TAF subjects despite the expected points in ITAF and TAF being the same. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 misrepresentation under the bonus scheme than under the bonus-cum-penalty scheme. Despite penalty provisions in incentive contracts being uncommon in practice, negative consequences are still attached to poor performance although they are not necessarily monetary in the short term. For example, managers may get fired if they are delivering bad or inconsistent performance. Dismissal would ultimately result in a loss of income for the manager. Managers may also lower their probability for promotion, which ultimately results in a loss of potential increase in income and prestige from promotion. Managers may also create a bad reputation for themselves that may adversely affect their future labor market. Given these real-world negative consequences associated with poor performance, I incorporated negative consequences into the compensation schemes in the experiment by deducting points from the subjects when they did not achieve their targets. On the other hand, points were awarded to subjects for achievement above the target, simulating bonuses for good performance. Prospect theory suggests that positive points for achievement above target would be perceived differently from negative points for achievement below target (Benzion, Rapoport, & Yagil, 1989; Kahneman & Tversky, 1979; Luft, 1994; Thaler, 1981). Under prospect theory, the increase in utility from the points reward for a 1% achievement above target would be less than the decrease in utility from the points deduction for a 1% achievement below target. However, since I awarded points for 1 5 The total expected points was -1,480,402 points for Conditions 4 and 8, and -1,536,022 points for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 2 above-target performance and deducted points for below-target performance in a similar manner across all eight conditions in my study, any results would be due to differences in the eight conditions rather than to the difference in utility from bonus versus penalty points. Description o f the Four Levels o f PMS PMS1: Reward SFM alone (first level). Subjects were rewarded 100% based on ROI performance in each period (ROIt). 2000 points were awarded and deducted for every percent of achievement above and below 80% of planned ROIt, respectively. Subjects were told that the planned ROIt was 39%. Feedback on gross operating income, ROI, and points reward was provided for the current period and all prior periods in a division and for all prior divisions. No information on the VDs was provided. PMS2: Reward SFM alone and provide VD information (second level). Subjects were rewarded in the same way as subjects in PMS1. Subjects were told that planned ROIt was 39%, planned R&DPt was 46 products, and planned P&EOEt was a 92 rating. Subjects were given similar feedback as subjects in PMS1 and, in addition, they were given feedback on the VDs. In PMS2, PMS3, and PMS4, subjects were informed that “the R&D productivity and P&E operating efficiency are important information that will be useful for understanding the relationships between R&D and P&E investments and gross operating income and improving [their] performance.” This mitigated the Conditions 1, 2, 3, 5, 6, and 7. The differences were due to rounding errors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 3 possibility that subjects might not attend to the VD information in the belief that they were irrelevant. PMS3: Reward SFM alone and provide VD information and VD weights (third level). Subjects were rewarded in the same way as subjects in PMS1 and PMS2. As in PMS2, subjects were told that planned ROIt was 39%, planned R&DPt was 46 products, and planned P&EOEt was a 92 rating. Subjects were given similar feedback as subjects in PMS2 and, in addition, they were informed that the company believed that the performance of each division was based 43.5% on ROI, 43.5% on the more important VD and 13% on the less important VD. The relative weights of the VDs were in the same proportion as the ratio of R&D and P&E investments for optimal T c t in each period. In other words, VDs were accurately weighted. MCPL literature has shown that people tend to integrate information additively and that they perceive the importance of a cue as the slope of the function relating the criterion to the cue rather than the proportion of the variance in the criterion explained by the cue (Brehmer & Qvamstorm, 1976). This supports the usage of VD weights as reflecting the slope of the function relating gross operating income to VDs rather than the proportion of the variance in gross operating income explained by VDs. PMS4: Reward SFM and accurately-weighted VDs (fourth level). Subjects were rewarded 43.5% based on ROI performance (ROIt), 43.5% based on the more important VD (R&DPt or P&EOEt), and 13% based on the less important VD Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 4 (R&DPt or P&EOEt) in each period.1 6 1850 points were awarded and deducted for every percent of achievement above and below 80% of planned ROIt, respectively. 1850 points were awarded and deducted for every percent of achievement above and below 80% of planned performance on the more important VD, respectively. 550 points were awarded and deducted for every percent of achievement above and below 80% of planned performance on the less important VD, respectively. As in PMS2 and PMS3, subjects were told that planned ROIt was 39%, planned R&DPt was 46 products, and planned P&EOEt was a 92 rating. Subjects were given the same feedback as subjects in PMS2 and PMS3. As in PMS3, subjects were also informed that the company believed that the performance of each division was based 43.5% on ROI, 43.5% on the more important VD, and 13% on the less important VD. Each VD was accurately weighted in PMS4 in the same proportion as the ratio of R&D and P&E investments for optimal in each period. Operationalizations of Dependent Variables Performance The performance of managers was operationalized as firm profits and not the performance on those measures that managers were being evaluated on in the PMS. 1 6 The Hackett Group Best Practices Benchmark Study of Planning and Performance Measurement surveyed companies that ranged from $15 million to $150 billion in annual revenues, although the exact number of companies was not disclosed (The Hackett Group, 1998). Almost half o f the companies in their survey used both operational and financial measures in their incentive schemes. For companies that were using both nonfinancial and financial measures, almost three quarters o f the weighting was still on financial measures. According to the AICPA/Maisel Survey, financial measures were given more importance than nonfinancial measures although there was no information on the relative weighting (AICPA & Maisel, 2001). This supports the continued weighting on ROI in PMS4. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 5 Since there were no other factors that affected firm profits other than the investment choices of the manager in the experiment, firm profits was the performance of the manager. In reality, firm profits are affected by many other variables outside the manager’s direct control such as external economic events, the actions of other divisions, and the actions of a manager’s supervisor. Performance in each period was operationalized as the gross operating income, 7 it. 7 tt captured the real economic effects of R&D and P&E investments on firm profits in the experiment since it disregarded the effects of different accounting treatments of R&D investment (immediate expensing) and P&E investment (capitalizing and depreciating). Performance of each manager was the mean rct over the total number of periods across all four divisions that were determined by the investment choices of the manager. The first three periods in each division were excluded in the performance because rc t in the first three periods were not determined 1 7 by the subject’s decisions but by prior expenditures set by the experimenter. Larger nt and mean 7tt indicated better performance. Performance = 1 °f d m sio n k % t j(T otal periods in 4 divisions -1 2 ). (5) 1 7 R&D and P&E investments affected 7 t, with a three-period lag. Hence, 7 r t in the first three periods of a division were determined by the R&D and P&E investments set by the experimenter in the prior periods before the subject begins investing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 Learning Learning is usually inferred from performance improvement induced by feedback over time (Goodman, 1998). The experiment design enables performance effects that are a result of differences in the types of feedback to be isolated from performance effects due to differences in incentives. As discussed earlier, performance differences across conditions with different feedback but identical incentives could be interpreted as learning effects. Performance differences across conditions with similar feedback but different incentives could be due to either a learning effect from paying more attention to information that is now rewarded or a pure motivation effect from following directions better in the incentive scheme. To further delineate learning and motivation effects, subjects were asked eight questions in the post-experimental questionnaire to measure their knowledge of the relationships between investments and gross operating income. The knowledge questions are Q7 - Q14 in Screen 15: Post-experimental questionnaire in Appendix A. These questions tested the subject’s knowledge about the relative importance of R&D versus P&E investments, the lagged relations between investments and gross operating income, the sign of the relationship between investments and gross operating income, and the nonlinear relationship between investments and gross operating income. Learning was operationalized as the number of correct answers to these knowledge questions. This is a post-test only measure. A pre-test of knowledge is not possible since it may prompt subjects before they begin the study, making the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 7 results difficult to generalize to situations where the pre-test has not been administered. Optimality o f Resource Allocation Optimality of resource allocation was operationalized as the mean squared difference (MSB) between the actual investments and the optimal investments over all periods. MSB between R&Dt and optimal R&Dt is equivalent to MSB between P&Et and optimal P&Et since R&Dt = 100 - P&Et and optimal R&Dt = 100 - Optimal P&Et. Smaller MSB indicated more optimal resource allocation. Appendix F shows the calculations of the MSB between the actual and optimal investments for TAF and ITAF. MSB between R&Dt and optimal R&Dt ~ X L X r r “ k(R&Dt ~ Optimal R&D t )2 / t otal periods in 4 divisions. (6) Attention to VB Information In the post-experimental questionnaire, subjects were asked Q3 to measure their attention to VD information (see Condition 2, Screen 15: Post-experimental questionnaire in Appendix A). Attention to VD information was operationalized as the total percentage of attention assigned by the subject to the two VDs performance measures: R&D productivity and P&E operating efficiency. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 Attention to the More Important Investment In the post-experimental questionnaire, subjects were asked Q4 to measure their attention to the effects of the more important investment on performance, which was operationalized as the percentage of attention assigned by subjects to the effects of the more important investment on performance (see Condition 1, Screen 15: Post- experimental questionnaire in Appendix A). P&E investment is the more important investment for TAF while R&D investment is the more important investment for ITAF. Attention to the Long-Term Impact o f Investments In the post-experimental questionnaire, subjects were asked questions Q5 and Q6 to measure their attention to the short-term versus long-term impact of R&D and P&E investments on performance (see Condition 1, Screen 15: Post-experimental questionnaire in Appendix A). Attention to the short-term effects of R&D investments on performance was operationalized as the points assigned to Item 5.1 (Current R&D investment decreases current ROI because current R&D investment is immediately expensed) and Item 5.2 (Current R&D investment increases current gross operating income and current ROI because current R&D productivity increases as new products are introduced). Attention to the long-term effects of R&D investments on performance was operationalized as the points assigned to Item 5.3 (Current R&D investment increases future gross operating income and future ROI because current R&D productivity increases as new products are introduced. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 9 However, increased current R&D productivity takes some time before it positively affects gross operating income and ROI). Attention to the short-term effects of P&E investments on performance was operationalized as the points assigned to Item 6.1 (Current P&E investment decreases current ROI because current P&E investment is capitalized and depreciated over 5 periods) and Item 6.3 (Current P&E investment increases current gross operating income and current ROI because current P&E operating efficiency increases with new and better plant and machinery). Attention to the long-term effects of P&E investments on performance was operationalized as the points assigned to Item 6.2 (Current P&E investment decreases future ROI because current P&E investment is capitalized and depreciated over 5 periods) and Item 6.4 (Current P&E investment increases future gross operating income and future ROI because current P&E operating efficiency increases with new and better plant and machinery. However, increased current operating efficiency takes some time before it positively affects gross operating income and ROI). Items 5.4 and 6.5 were open items in which subjects described other ways in which they thought about the effects of R&D and P&E investments on their performance. Items 5.4 and 6.5 that were weighted were content analyzed independently by two coders into three categories: short-term effect, long-term effect, and neither short-term nor long-term effect. Inter-coder reliability of 0.74 was calculated using Rrippendorff s (1980) alpha. Disagreements between coders were resolved after calculating the inter-coder reliability. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 0 Hence, attention to the short-term effects of investments on performance was operationalized as the total points assigned by the subject to Items 5.1, 5.2, 6.1, and 6.3 as well as Items 5.4 and 6.5 that were categorized as short-term effects in the content analysis. Attention to the long-term effects of investments on performance was operationalized as the total points assigned by the subject to Items 5.3, 6.2, and 6.4 as well as Items 5.4 and 6.5 that were categorized as long-term effects in the content analysis. Heuristics Used in Learning In the post-experimental questionnaire, subjects were asked Q15 to measure the relative usage of various heuristics used for understanding the relationship between investments and performance (see Condition 1, Screen 15: Post- experimental questionnaire in Appendix A). In the post-experimental questionnaire, Item 15.1 represented a difference- based heuristic based on a locational criteria: adjacent-rows differences heuristic. Item 15.2 represented an exemplar-based heuristic based on an informational criteria: best-exemplars heuristic. Item 15.3 represented a chunk-based heuristic based on a locational criteria: recent versus initial prototypes heuristic. Item 15.4 represented a chunk-based heuristic based on a locational criteria and adapted for time lags: recent vs initial prototypes heuristic adapted for time lags. Subjects were also asked to describe any other strategies that were not given as one of the options in Item 15.5. Item 15.5 that was weighted was content-analyzed independently by two coders into 13 categories. Of the 13 categories used in the content analysis, 12 represented the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 heuristics identified by Hutchinson and Alba (1997) and the last category was for no identifiable heuristic (see Table 1). Inter-coder reliability of 0.69 was calculated using Krippendorff s (1980) alpha. Disagreements between coders were resolved after calculating the inter-coder reliability. In assessing the relative attention of subjects to various items and the relative usage of heuristics, subjects were asked to divide 100 points across the various items such that a higher rating indicated a more important item and the sum of the ratings was 100. Cook and Stewart (1975) noted that using such a subjective method to obtain descriptions of a subject’s judgmental policy corresponded reasonably well to a judgmental policy description obtained by statistical regression. Using the weights obtained by this subjective method also predicted actual judgments better than using equal weights on all cues. Manipulation Checks In the post-experimental questionnaire, the subject was asked Q1 to assess whether the first factor, Firm (TAF versus ITAF), was effectively manipulated (see Condition 1, Screen 15: Post-experimental questionnaire in Appendix A). Subjects were asked which type of investment (R&D or P&E) they should invest more to maximize the performance of a division across all periods. To assess whether the second factor, PMS, was effectively manipulated, the subject was asked Q2 (see Condition 1, Screen 15: Post-experimental questionnaire in Appendix A). Subjects were asked the relative weights that were explicitly placed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 2 on achieving results in ROI, R&D productivity, and P&E operating efficiency in their compensation scheme in the experiment. Quality of Data from Web-Based Experiments There may be concern over the quality of data that is collected on the web compared with data collected in a traditional laboratory. The number of psychologically related web studies has increased significantly, indicating the increasing acceptance of web studies. As of December 3, 2002, there were 126 experiments on the Internet in the American Psychological Society list of Psychological Research on the Net maintained by Krantz (2002). The general conclusion of studies comparing Web research and laboratory research on the same topic is that the two research methods yield the same results (e.g., Bimbaum, 1999, 2000; Buchanan and Smith, 1999; Krantz, Ballard, & Scher, 1997; Pasveer and Ellard, 1998; Pettit, 1999; Stanton, 1998). In a web-based experiment, there is limited control over the identity of the person who is undertaking the experiment. Various measures were taken to authenticate the person participating in the experiment. The call for participation was made only to MBA students during class and via email. Interested participants were asked to contact the experimenter via email. The experimenter issued a unique log-in name and password to a person only after verifying that the person was a MBA student registered with the Marshall School of Business. Only persons with issued log-in names and passwords could participate in the experiment. Potential subjects logging on to the website were instructed to undergo the experiment alone. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 3 A subject may also attempt to participate in the experiment multiple times. Various measures were taken to prevent this. Subjects were instructed to participate in the experiment only once. Each subject logged on to participate in the experiment with a unique name and assigned password. The computer program was coded such that once a subject completed the experiment, the subject would not be able to log on to the website again. Since the source code for HTML (Hypertext Markup Language) pages is accessible to anyone, subjects may examine the source code to try to identify the optimal strategy for achieving the maximum reward. As such, the source code for the equations linking investments to nt, ROIt, R&DPt, and P&EOEt was made inaccessible to the subject. The back button function built in all web browsers were also disabled to prevent subjects from examining prior web pages and submitting them again, possibly altering their data. Random Heterogeneity of Subjects Subjects were randomly assigned to experimental conditions to control the random heterogeneity of subjects. I also recruited only MBA students for a more homogenous demographic background. In addition, several demographic characteristics of subjects were measured and the results were analyzed to ensure no systematic biases across the eight experimental conditions (see Condition 1, Screen 6: Demographic Questionnaire in Appendix A). The demographic variables measured include gender, age, number of years of full-time working experience, major functional responsibility in work experience (accounting, auditing or taxation; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 4 finance, banking or investing; sales & marketing; general management; human resources; engineering & operations; R&D), number of management accounting courses taken for MBA degree to date, and number of financial accounting courses taken for MBA degree to date. Standardization of Experimental Conditions The administration of each experimental condition, such as the instructions given and the information displayed, was standardized where possible. However, using a web-based delivery of the experimental instrument introduced various difficulties in standardizing all aspects of the administration of the experiment. For example, I could not control and standardize the environment in which the subject participated in the experiment such as the time, room, and presence of external distractions. Despite the potential increase in error variance, significant and interesting results were obtained. Collection of Data over 9 Months The data was collected over a period of 9 months from June 2002 to February 2003. Thirty-five subjects participated in the experiment from June, 2002 to July, 2002. Sixty-seven subjects participated from September, 2002 to November, 2002. Sixty-one subjects participated from December, 2002 to February, 2003. The results from these three groups of subjects were analyzed separately. There were no differences in the direction of the results in the tests of hypotheses across the three groups of subjects. Only the p- values in the tests of hypotheses differ across the three Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 groups of subjects. As such, the three groups of subjects were combined and analyzed together. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 6 CHAPTER 4: DATA ANALYSIS ANOVA and ANCOVA The data was analyzed by performing ANOVA and ANCOVA on the various dependent variables with the two independent variables, Firm and PMS. The emphasis of the ANOVA and ANCOVA procedures was to determine the significance of the interaction effects between Firm and PMS. ANOVA and ANCOVA assume that the error term in the model is independent and normally distributed with a constant error variance. To ensure that the error term was independent, subjects were randomly assigned to the conditions in the experiment (Kirk, 1995). The F-test is quite robust to moderate violations of normality (Lindquist, 1953 as cited in Kirk, 1995) but is more sensitive to violations of the homogeneity of variance (Wilcox, 1987 as cited in Kirk, 1995). Normality and homogeneity of variances were tested and nonparametric alternatives to the F-test in ANOVA and ANCOVA were used where the assumptions were violated. Multiple Comparison Tests ANOVA and ANCOVA analyses only indicate whether there are any significant differences among the dependent variables, but do not indicate specifically which of the means differ. To test the individual hypotheses, 10 multiple comparison tests were conducted (Table 4). To control for type I error when multiple contrasts were performed, I used the Holm’s test and reported the adjusted ^-values (Holm, 1979). Kirk (1995) recommended Holm’s test as a more powerful test to control the type I error when a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 7 specific number of a priori contrasts, C, are tested. To increase the power of the test, Holm’s test modifies the Bonferroni single-step procedure into a step-down 1 Is-I procedure and uses the multiplicative inequality, 1 - (1 - a) , rather than the additive inequality, a/C. Under Holm’s test, the test statistics are ranked on the basis of the /^-values of the test statistics from the largest to the smallest. The largest test statistic is tested at the 1 - (1 - a )1/c level of significance, the second largest at the 1 - (1 - a)1 /(C ” level of significance, the third largest at the 1- (1 - a )1/(C' 2 ) level of significance, and so forth. The test stops at a nonsignificant test statistic. For the 10 multiple contrasts in this experiment, the critical two-tailed /(-values to control type I error rate at a = .05 and a = .10 are in Table 5. Table 4 Description of Multiple Contrasts Contrast Common factor Description of contrast 1 . Condition 1 vs. Condition 5 PMS1 TAF vs. ITAF 2. Condition 2 vs. Condition 6 PMS2 TAF vs. ITAF 3. Condition 3 vs. Condition 7 PMS 3 TAF vs. ITAF 4. Condition 4 vs. Condition 8 PMS4 TAF vs. ITAF 5. Condition 1 vs. Condition 2 TAF PMS1 vs. PMS2 6. Condition 5 vs. Condition 6 ITAF PMS1 vs. PMS2 7. Condition 2 vs. Condition 3 TAF PMS2 vs. PMS3 8. Condition 6 vs. Condition 7 ITAF PMS2 vs. PMS3 9. Condition 3 vs. Condition 4 TAF PMS3 vs. PMS4 10. Condition 7 vs. Condition 8 ITAF PMS3 vs. PMS4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 Table 5 Critical Two-tailed p-values under Holm’s Test for 10 Multiple Contrasts Test statistic Critical p-value a = .05 a = .10 Largest test statistic 1 -(1 -a )1 /1 0 =.0051 l > — * l s o l i © o L ft 2n d largest test statistic 1 -(1 -a )1 7 9 =.0057 t i £ s o 1 1 © > — « * 3rd largest test statistic 1 -(1 -a )1 /8 =.0064 -(1 - a)1 7 8 = .0131 4th largest test statistic 1 -(1 - a)1 7 7 =.0073 - (1 - a)1 7 7 = .0149 5th largest test statistic © 1 1 i i V — H < 1 £ a s 1 1 O * 6th largest test statistic 1 - (1 - a)1 7 5 = .0102 - (1 - a)1 7 5 = .0209 7th largest test statistic 1 - (1 - a)1 7 4 = .0127 - (1 - a)'7 4 = .0260 8th largest test statistic 1 -(1 - a)1 7 3 = .0170 - (1 - a)1 7 3 = .0345 9th largest test statistic 1 - (1 - a)1 7 2 = .0253 -(1 - a )1 7 2 = .0513 Smallest test statistic 1 - (1 - a) = .05 -(1 - a) = .10 Manipulation Checks The manipulation checks indicated that the two independent variables, Firm and PMS, were successfully manipulated. In a post-experimental question asking subjects to choose the type of investment that they should invest more to maximize the performance of a division across all periods, TAF subjects (83.42%) selected P&E investment more often than ITAF subjects (52.85%), t = 4.49, p < .01 (one tailed).1 8 Hence, Firm was successfully manipulated since TAF subjects recognized 1 8 The Satterthwaite t-test was used because the equality of variances test rejected the assumption of equal variances (Folded F = 1.79, p < .01). The nonparametric Wilcoxon two-sample test also supports the f-test results, Wilcoxon Z = 4.02, p < .01 (one-tailed). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 9 that P&E investment was more important for firm performance more often than ITAF subjects did. In a post-experimental question asking subjects the explicit weights (out of 100%) placed on achieving results in ROI, R&D productivity, and P&E operating efficiency in their compensation schemes, the weights perceived by subjects were reflective of the actual weights in the PMS that they were assigned to, indicating that PMS was successfully manipulated. The weights placed on ROI by subjects in PMS1 (M ~ 65.86%, SD = 27.29%), PMS2 (M = 65.00%, SD = 32.30%), and PMS3 (M = 61.61%, SD = 30.10%) were not significantly different from each other, t = 0.48, adjusted p - .63 (two-tailed). The ROI weight of subjects in PMS4 (M = 41.39%, SD - 19.75%), however, was significantly smaller than the ROI weight of subjects in PMS1, PMS2, and PMS3, t = 4.59, adjusted p - < .01 (one-tailed). From the viewpoint of the weight placed on ROI as perceived by subjects, PMS was successfully manipulated since PMS4 placed a smaller weight on ROI (43.5%) than the other PMSs (100%). The weights placed on R&D productivity by subjects in PMS1 (M = 16.51%, SD = 12.80%), PMS2 (M = 13.00%, SD = 14.84%), and PMS3 (M= 15.93%, SD = 15.37%) were not significantly different, t = 0.76, adjusted p = .45 (two-tailed). The R&D productivity weight of subjects in PMS4 (M = 25.93%, SD - 12.51%), however, was significantly larger than the R&D productivity weight of subjects in PMS1, PMS2, and PMS3, t = 4.32, adjustedp < .01 (one-tailed). From the viewpoint of the weight placed on R&D productivity as perceived by subjects, PMS was Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 successfully manipulated since PMS4 placed larger weight on R&D productivity (13% in TAF and 43.5% in ITAF) than the other PMSs (0%). The weights placed on P&E operating efficiency by subjects in PMS1 (M = 17.63%, SD = 18.47%), PMS2 (M = 22.00%, SD = 26.65%), and PMS3 (M = 22.54%, SD = 22.19%) were not significantly different, t = 1.11, adjusted p - .27 (two-tailed). The P&E operating efficiency weight of subjects in PMS4 (M = 32.85%, SD = 18.15%) was significantly larger than the P&E operating efficiency weight of subjects in PMS1, PMS2, and PMS3, t = 3.14, adjusted p < .01 (one tailed). From the viewpoint of the weight placed on P&E operating efficiency as perceived by subjects, PMS was successfully manipulated since PMS4 placed larger weight on P&E operating efficiency (43.5% in TAF and 13% in ITAF) than the other PMSs (0%). The weights placed on P&E operating efficiency by subjects in the TAF and PMS4 condition (M = 44.37%, SD = 16.44%) were significantly larger than those of subjects in the ITAF and PMS4 condition (M ~ 21.33%, SD = 11.24%), t = 3.62, p < .01 (one-tailed). The weights placed on R&D productivity by subjects in the TAF and PMS4 condition (M = 20.63%, SD = 11.00%) were significantly smaller than that of subjects in the ITAF and PMS4 condition (M = 31.24%, SD = 11.87%), t = - 2.52, p < .01 (one-tailed). The weights placed on ROI by subjects in TAF and PMS4 condition (M - 35.11%, SD - 21.75%) were not significantly different from those of subjects in ITAF and PMS4 condition (M= 47.67%, SD = 15.60%), t = 1.47, p - .14 (two-tailed). This suggested that PMS was successfully manipulated since PMS4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 placed larger weight on P&E operating efficiency in TAF (43.5%) than ITAF (13%); smaller weight on R&D productivity in TAF (13%) than ITAF (43.5%); and same weight on ROI (43.5%) in TAF as in ITAF. Tests of Hypotheses HI - H4: Performance ANOVA for mean nt for each subject indicated that managerial performance was significantly different across conditions, F = 13.47, p < .01 (Table 6). The interaction effect, Firm X PMS, was significant, indicating that managerial performance under different PMS was dependent on the type of firm, F = 2.93, p = .04. Table 6 ANOVA for Performance Source SS df MS F P Model 7530.54 7 1075.79 13.47** < .01 Error 12375.29 155 79.84 Corrected Total 19905.83 162 Firm 6154.31 1 6154.31 77.08** <.01 PMS 678.76 3 226.25 2.83* .04 Firm X PMS 700.92 3 233.64 2.93* .04 Note. R2= .38 for Model *p < .05, **p < .01 The Kolmogorov-Smimov test of normality indicated that Condition 5 (D = 0.14, p > .15) and Condition 8 (D = 0.12, p > .15) did not violate the normality assumption; Condition 1 (D = 0.19, p = .06) and Condition 2 (D = 0.18, p = .06) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 violated the normality assumption at a = .10; while Condition 3 (D = 0.20, p = .04), Condition 4 (D = 0.20, p = .02), Condition 6 (Z) = 0.22, p = .02), and Condition 7 (D = 0.25, p < .01) violated the normality assumption at oc = .05. The Kruskal-Wallis test of performance differences across conditions, equivalent to a parametric one way ANOVA, rejected the hypothesis that performance was identical across all conditions, x2 ~ 75.19, p < .01. The Levene’s (1960) test indicated that the variances in performance across conditions were not homogeneous, F = 6.80, p < .01. The Welch’s variance-weighted ANOVA (Welch, 1951), recommended when variances were not homogeneous, rejected the hypothesis that performance was equal across conditions, F = 12.34, p < .01. Attempts at transformations of the performance variable did not remove the heterogeneity of variances. 1 9 However, the results were robust since all the nonparametric analyses supported the ANOVA analysis. Multiple contrasts were conducted to test the individual hypotheses. Table 7 and Figure 4 display the results for H1-H4. Table 7 Performance (in $ million) by Condition Firm P M S 1 P M S 2 P M S 3 P M S 4 TAF 1 3 4 .3 3 a 1 3 2 .7 9 a 1 3 4 .1 4 , 1 3 4 .0 0 a (2 .8 3 ) (5 .3 8 ) (2 .8 5 ) (3 .0 9 ) ITAF 116.81b 1 2 0 .6 2 b 120.64b 1 2 7 .9 8 a (1 4 .3 1 ) (1 4 .0 9 ) (1 2 .4 6 ) (5.16) Note. Values enclosed in parentheses represent standard deviation. Means with different subscripts differ significantly at adjustedp < .05 using the Holm’s test. 1 9 Log, inverse, and square root transformations were attempted. ANOVA and multiple comparison results with transformed variables were similar to the results obtained with untransformed variables. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 140 n 13 5 - - A - I T A F ^ 120 1 15 - 110 P M S 1 P M S 2 P M S 3 P M S 4 Figure 4. Performance (in $ million) by condition HI: When rewarded with SFM alone, TAF managers perform better than ITAF managers. When rewarded with SFM alone (PMS1), TAF managers (M = $134.33 million, SD = $2.83 million) performed significantly better than ITAF managers (M - $116.81 m illion, SD = $14.31 million), t - 6.19, adjusted p < .01 (one-tailed), Wilcoxon Z = 4.55, p < .01 (one-tailed). HI was supported. The performance of TAF managers was also less variable than the performance of ITAF managers under PMS1, Folded F = 25.65,p < .01 (two-tailed). Additional contrasts were conducted to assess whether the performance gap between managers of TAF and ITAF decreased with different PMSs. When VD information was added (PMS2), TAF managers (M - $132.79 million, SD = $5.38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 4 million) still performed significantly better than ITAF managers (M = $120.62 million, SD = $14.09 million), t = 4.30, adjusted p < .01 (one-tailed), Wilcoxon Z = 3.44, p < .01 (one-tailed). The performance of TAF managers was less variable than the performance of ITAF managers under PMS2, Folded F = 6 .8 6 , p < .01 (two- tailed). When VD weights were added (PMS3), TAF managers (M = $134.14 million, SD = $2.85 million) still significantly outperformed ITAF managers (M = $120.64 million, SD - $12.46 million), t = 4.84, adjusted p < .01 (one-tailed), Wilcoxon Z = 4.99, p < .01 (one-tailed). The performance of TAF managers was less variable than the performance of ITAF managers under PMS3, Folded F = 19.08,/) < .01 (two-tailed). Finally, when VDs were rewarded (PMS4), TAF managers (M - $134.00 million, SD - $3.09 million) only performed marginally better than ITAF managers (M = $127.98 million, SD = $5.16 million), t - 2.18, adjustedp = .09 (one tailed), Wilcoxon Z = 3.80, p < .01 (one-tailed). While the performance of TAF managers was still statistically less variable than the performance of ITAF managers under PMS4 (Folded F = 2.79, p = .03), the variance in performance of ITAF managers under PMS4 was smaller than the variance of ITAF managers under PMS1, PMS2, and PMS3. Hence, the performance gap between TAF managers and ITAF managers diminished only when VDs were rewarded in the PMS. The performance of ITAF managers was also less variable when VDs were rewarded in the PMS as compared to when VDs were not rewarded. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 H2 (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their performance more than TAF managers do. When VD information was provided as compared to when it was not, both TAF managers (t = -0.55, adjustedp = 1.00 [one-tailed], Wilcoxon Z = -0.43 p - .33 [one-tailed]) and ITAF managers (t = 1.35, adjustedp = .45 [one-tailed], Wilcoxon Z = 0.92, p = .18 [one-tailed]) did not improve their performance significantly. H2 was not supported. H3 (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their performance more than TAF managers do. When VD weights were provided as compared to when they were not, both TAF managers (t = 0.48, adjusted p = 1.00 [one-tailed], Wilcoxon Z = 0.30, p = .38 [one-tailed]) and ITAF managers (t - 0.01, adjusted p = 1.00 [one-tailed], Wilcoxon Z = 0.43, p = .33 [one-tailed]) did not improve their performance significantly. H3 was not supported. H4 (Effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their performance more than TAF managers do. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 When VDs were rewarded as compared to when they were not, TAF managers did not improve their performance significantly (t = -0.05, adjusted p — 1.00 [one-tailed], Wilcoxon Z = -0.17, p = .43 [one-tailed]), but ITAF managers improved their performance significantly (t = 2.66, adjusted p = .03 [one-tailed], Wilcoxon Z - 1.89, p = .03 [one-tailed]). H4 was supported. The results for the multiple comparisons of performance suggested that the provision of VD information and VD weights did not significantly improve managerial performance in both TAF and ITAF. This was despite the fact that subjects in these conditions were explicitly told that VDs were important information that was useful in understanding the relationships between their investments and performance. Rewarding VDs improved the performance of ITAF managers but not that of TAF managers. Hence, VDs were more useful for improving performance in ITAF than in TAF, but only when VDs were explicitly rewarded. The variability of managerial performance was lower in TAF than ITAF. When VDs were explicitly rewarded, the variance in the performance of ITAF managers was smaller as compared to the variance under other PMS. H5 - H8: Learning ANOVA for the mean number of correct answers to the post-experimental knowledge questions indicated that the learning of managers was significantly different across conditions, F = 4.97, p < .01 (Table 8 ). The interaction effect, Firm X PMS, was not significant, F = 1.84, p = .14. Only the main effect for Firm was significant, F - 24.54, p < .01. Learning was significantly higher for TAF managers Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 7 (M = 4.07, SD = 1.25) than ITAF managers (M = 3.15, SD = 1.16), t = 4.90,/? < .01 (one-tailed), and learning was not moderated by the type ofPMS. Table 8 ANOVA for Learning Source SS Df MS F P Model 49.57 1 7.08 4.97** < . 0 1 Error 220.64 155 1.42 Corrected Total 270.20 162 Firm 34.94 1 34.94 24.54** < . 0 1 PMS 6 . 2 1 3 2.07 1.45 .23 Firm X PMS 7.86 3 2.62 1.84 .14 Note. R2 = . 18 for Model *p < .05, **/?<.01 The Kolmogorov-Smimov test of normality indicated that Condition 7 (D = 0 .1 6 , p - .1 4 ) and Condition 8 (D - 0 .1 7 , p = .1 3 ) did not violate the normality assumption; Condition 4 (D = 0 .1 9 , p - .0 6 ) violated the normality assumption at a = .1 0 ; and Condition 1 (D = 0 .2 2 , p = .0 2 ) , Condition 2 (D = 0 .2 1 ,/ ? = .0 2 ) , Condition 3 (D = 0 .2 5 ,/ ? < .0 1 ) , Condition 5 (D = 0 . 2 4 ,p < .0 1 ) and Condition 6 (D = 0 .3 1 ,/ ? < .0 1 ) violated the normality assumption at a = .0 5 . A Kruskal-Wallis test rejected the hypothesis that learning was identical across all conditions, % 2 - 2 9 .7 8 , p < .0 1 . The Levene’s ( 1 9 6 0 ) test indicated that the variances in learning across conditions were homogeneous, F = 1 .0 2 , p = .4 2 . The results were robust since the nonparametric analyses supported the ANOVA analysis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 Multiple contrasts were then conducted to test the individual hypotheses. Table 9 and Figure 5 display the results for H5 - H8 . Table 9 Learning by Condition Firm P M S 1 P M S 2 P M S 3 P M S 4 TAF 4 .0 5 a 4 .7 1 a 3.85b 3 .6 7 b (1 .2 7 ) (1 .1 5 ) ( 1 .2 7 ) (1 .1 5 ) ITAF 3 .1 9 b 3 .0 5 b 3 .3 3 b 3 .0 0 b (1 .1 2 ) (1 .0 3 ) (0 .8 9 ) (1 .5 5 ) Note. Values enclosed in parentheses represent standard deviation. Means with different subscripts differ significantly at adjustedp < .10 using the Holm’s test. 5 -| 4.5 - $ 4 - I 3 3.5- 3 - 2.5 - PMS1 PMS2 PMS3 PMS4 Figure 5. Learning by condition H5: When rewarded with SFM alone, TAF managers learn more about the relationships between investments and firm profits than ITAF managers. ■TAF ■ITAF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 9 When rewarded with SFM alone (PMS1), TAF managers (M = 4.05, SD = 1.27) learned better than ITAF managers (M = 3.19, SD = 1.12), t = 2.28, adjustedp = .10 (one-tailed), Wilcoxon Z - 1.93, p = .03 (one-tailed). H5 was supported, corresponding to the results for HI. Additional contrasts were conducted to assess whether the learning difference between managers of TAF and ITAF would change under different PMS. When VD information was added (PMS2), TAF managers (M = 4.71, SD = 1.15) still learned significantly better than ITAF managers (M= 3.05, SD = 1.03), t = 4.40, adjustedp < .01 (one-tailed), Wilcoxon Z = 4.10, p < .01 (one-tailed). When VD weights were added (PMS3), TAF managers (M = 3.85, SD = 1.27) no longer learned better than ITAF (M - 3.33, SD = 0.89), t = 1.39, adjusted p ~ .42 (one-tailed), Wilcoxon Z = 1.63, p — .05 (one-tailed). Finally, when VDs were rewarded (PMS4), TAF managers (M== 3.67, SD = 1.15) also did not learn better than ITAF managers (M = 3.00, SD = 1.55), t - 1.81, adjustedp = .25 (one-tailed), Wilcoxon Z = 1.60,p = .06 (one-tailed). Hence, when VD weights were added and when VDs were rewarded, the learning difference between TAF and ITAF managers disappeared. When VD weights were added, TAF managers still outperformed ITAF managers despite their learning being the same. It was only when VDs were rewarded that the performance of ITAF managers caught up with that of TAF managers. It could be argued that when VD weights were added, TAF managers in Condition 3 are still be better motivated than ITAF managers in Condition 7 since only SFM was rewarded even though VD information and VD weights were Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 0 available. Rewarding only SFM would motivate managers to allocate more resources to tangible assets than intangible assets. When VDs were rewarded, the motivation of ITAF managers in Condition 8 improved vis-a-vis Condition 7 more than the motivation of TAF managers in Condition 4 improved vis-a-vis Condition 3. ITAF managers who were now rewarded based on VDs would be motivated to allocate more resources to intangible assets than when only SFM was rewarded. On the other hand, TAF managers who were now rewarded on VDs were not expected to be motivated to allocate more resources to tangible assets than when only SFM was rewarded. These results suggested that motivation rather than learning explained the difference in performance between ITAF and TAF managers when VD weights were added and explain the non-difference in performance between ITAF and TAF managers when VDs were rewarded. H6 a (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. When VD information was provided as compared to when it was not, both TAF managers (t = 1.75, adjusted p = .25 [one-tailed], Wilcoxon Z - 1.72, p = .04 [one-tailed]) and ITAF managers (t - -0.36, adjusted p - 1.00 [one-tailed], Wilcoxon Z = -0.82, p = .21 [one-tailed]) did not learn better. H6 a was not supported, corresponding to the results for H2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 H6 b (Attention to the more important investment from having YD information): When VD information is provided as compared to when it is not, managers pay more attention to the effects of the more important investment on performance. H6 c (Attention to the long-term impact of investments from having VD information): When VD information is provided as compared to when it is not, managers pay more attention to the long-term impact of investments on performance. When VD information was provided as compared to when it was not, both TAF managers (t = 1.26, adjusted p = 0.42 [one-tailed], Wilcoxon Z= 1.28, p = .10 [one-tailed]) and ITAF managers (t = -1.65, adjusted p - 0.25 [one-tailed], Wilcoxon Z = -1.91, p = .03 [one-tailed]) did not pay more attention to the effects of the more important investment on performance. H6 b was not supported. When VD information was provided as compared to when it was not, both TAF managers (t = 0.22, adjustedp = 1.00 [one-tailed], Wilcoxon Z = 0.14, p = .45 [one-tailed]) and ITAF managers (t = 0.06, adjustedp = 1.00 [one-tailed], Wilcoxon Z = 0.12, p = .45 [one-tailed]) did not pay more attention to the long-term impact of their investments on performance. H6 c was not supported. TAF and ITAF managers not paying more attention to the effects of the more important investment or the long-term impact of their investments on performance when VD information were provided might partially explain why their learning did Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 2 not improve. The latter is contrary to the argument of proponents of VDs that the addition of VDs would overcome myopia in managers. H7a (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. When VD weights were provided as compared to when they were not, TAF managers learn marginally worse (t = -2.32, adjusted p = .10 [one-tailed], Wilcoxon Z = -2.06, p = .02 [one-tailed]) and ITAF managers (t = 0.74, adjusted p = 1.00 [one tailed], Wilcoxon Z = 1.34, p = .09 [one-tailed]) did not learn better. H7a was not supported, corresponding to the results for H3. H7b (Attention to VD information from having VD weights): When VD weights are provided as compared to when they are not, managers pay more attention to VD information. H7c (Attention to the more important investment from having VD weights): When VD weights are provided as compared to when they are not, managers pay more attention to the effects of the more important investment on performance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 H7d (Attention to the long-term impact of investments from having YD weights): When VD weights are provided as compared to when they are not, managers pay more attention to the long-term impact of investments on performance. When VD weights were provided as compared to when they were not, TAF managers paid less attention to VD Information (t = -2.55, adjusted p = .02 [one tailed], Wilcoxon Z = -1.72, p = .04 [one-tailed]), while ITAF managers did not pay more attention to VD information (t ~ -0.58, adjustedp = .28 [one-tailed], Wilcoxon Z = -0.76,p = .22 [one-tailed]). H7b was not supported. When VD weights were provided as compared to when they were not, TAF managers did not pay more attention to the effects of the more important investment on performance (t = 0.41, adjusted p - 1.00 [one-tailed], Wilcoxon Z = 0.69, p - .24 [one-tailed]) while ITAF managers paid marginally more attention to the effects of the more important investment on performance (t = 2.24, adjusted p - .08 [one tailed], Wilcoxon Z = 2.55, p < .01 [one-tailed]). H7c was partially supported. When VD weights were provided as compared to when they were not, TAF managers paid less attention to the long-term impact of their investments on performance (t = -2.37, adjusted p = .05 [one-tailed], Wilcoxon Z = -1.73, p - .04 [one-tailed]) and ITAF managers did not pay more attention to the long-term impact of their investments on performance (t = -0.49, adjusted p - 1.00 [one-tailed], Wilcoxon Z = -0.14,/? = .45 [one-tailed]). H7d was not supported. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 4 TAF managers paying less attention to VD information and the long-term impact of their investments on performance and not paying more attention to the effects of the more important investment on performance when VD weights were provided might explain why their learning decreased marginally. ITAF managers did not improve in their learning when VD weights were added possibly because they did not pay more attention to VD information and the long-term impact of their investments on performance even though they paid marginally more attention to the effects of the more important investment on performance. H8 a (Learning effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their learning about the relationships between investments and firm profits more than TAF managers do. When VDs were rewarded as compared to when they were not, both TAF managers (t = -0.49, adjusted p = 1.00 [one-tailed], Wilcoxon Z = -0.56, p = .29 [one-tailed]) and ITAF managers (t = -0.91, adjustedp = 1.00 [one-tailed], Wilcoxon Z = 1.08, p = .14 [one-tailed]) did not learn better. H8 a was not supported, in contrast to the results for H4. Hence, the greater improvement in performance for ITAF managers compared to TAF managers in H4 when VDs were rewarded was not due to greater improvement in learning but was more likely due to better motivation of ITAF managers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 5 It is possible that when VD information and VD weights are provided but not rewarded, managers do not attend to the VD information, VD weights, and long-term impact of their investments on performance. As such, managers do not leam from having VD information and VD weights until the VDs are actually compensated. H8 b (Attention to VD information from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to VD information. H8 c (Attention to the more important investment from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to the effects of more important investment on performance. H8 d (Attention to the long-term impact of investments from rewarding VDs): When VDs are rewarded as compared to when they are not, managers pay more attention to the long-term impact of investments on performance. When VDs were rewarded as compared to when they were not, both TAF managers {/ = 2.61, adjusted p — .02 (one-tailed), Wilcoxon Z = 1.86,/? = .03 [one tailed]) and ITAF managers (t - 2.53, adjusted p = .02 [one-tailed], Wilcoxon Z = 2.41, p < .01 [one-tailed]), paid more attention to VD information. H8 b was supported. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 6 When VDs were rewarded as compared to when they were not, both TAF managers (t = 0.52, adjustedp = 1.00 [one-tailed], Wilcoxon Z - 0.27,/? = 0.39 [one tailed]) and ITAF managers (t = 0.68, adjusted p - 1.00 [one-tailed], Wilcoxon Z = 0.20, p = .42 [one-tailed]) did not pay more attention to the effects of the more important investment on performance. H 8 c was not supported. When VDs were rewarded as compared to when they were not, TAF managers paid more attention to the long-term impact of investments on performance (t = 2.85, adjusted p = .01 [one-tailed], Wilcoxon Z = 1.85, p = .03 [one-tailed]), while ITAF managers did not pay more attention to the long-term impact of investments on performance (t — -0.61, adjusted p = 1.00 [one-tailed], Wilcoxon Z - -0.59,/? = .28 [one-tailed]). H8 d was partially supported. The results in H 8 b-c supported the results in H 8 a. When VDs were rewarded, ITAF managers paid more attention to VD information but not to the effects of the more important investment and the long-term impact of their investments on performance. TAF managers also paid more attention to VD information and the long-term impact of their investments on performance but not to the effects of the more important investment on performance. As such, ITAF managers did not improve in their learning more than TAF managers. H9 - HI 2: Optimality o f Resource Allocation ANOVA for the MSB between the actual and optimal investments indicated that the optimality of the resource allocation was significantly different across conditions, F = 29.52, p < .01 (Table 10). In parallel to the ANOVA results on the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 7 performance variable, the interaction effect, Firm X PMS, was significant, F = 2.63, p = .05. The optimality of resource allocation in different PMS was dependent on the type of firm. Table 10 ANOVA for Optimality of Resource Allocation Source SS df MS F P Model 1034358263 1 14776552 29.52** < . 0 1 Error 77585354 155 500551 Corrected Total 181021216 162 Firm 94699266 1 94699266 189.19** < . 0 1 PMS 4591339 3 1530446 3.06* .03 Firm X PMS 3949811 3 1316604 2.63* .05 Note. R2 = . 16 for Model *p < .05, **p < .01 The Kolmogorov-Smimov test of normality indicated that, Condition 5 (D = 0.13, p > .15), Condition 6 (D = 0.14, p > .15), and Condition 8 (D = 0.13,p > .15) did not violate the normality assumption; Condition 3 (D = 0.19, p = .06) and Condition 7 (D = 0.19, p - .06) violate the normality assumption at a = .10; and Condition 1 (D — 0.22, p = .02), Condition 2 (D = 0.19,/? = .04), and Condition 4 (D = 0.25, p < .01) violated the normality assumption at a = .05. A Kruskal-Wallis test rejected the hypothesis that the optimality of resource allocation was identical across all conditions, y2 ~ 105.88, p < .01. The Levene’s (1960) test indicated that the variances across conditions were not homogeneous, F(7) = 10.21, p < .01. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 8 Welch’s variance-weighted ANOVA (Welch, 1951) rejected the hypothesis that the optimality of resource allocation was equal across conditions, F = 27.46, p < .01. Hence, the nonparametric analyses supported the ANOVA analysis. Multiple contrasts were then conducted to test the individual hypotheses. Table 11 and Figure 6 display the results for H5 - H8 . Table 11 Optimality of Resource Allocation by Condition Firm P M S 1 P M S 2 P M S 3 P M S 4 TAF 4 3 1 .4 4 a 3 8 9 .6 9 a 3 9 3 .3 0 a 389.96a (3 1 5 .1 4 ) (2 8 0 .4 7 ) (2 7 7 .1 9 ) (2 7 6 .5 9 ) ITAF 2 3 0 2 .6 7 b 1 994.32b 1 9 9 6 .2 2 b 1 4 1 4 .5 2 c (1 1 5 3 .7 9 ) (1 1 0 0 .2 3 ) (8 8 4 .8 9 ) (5 9 8 .3 3 ) Note. Values enclosed in parentheses represent standard deviation. Means with different subscripts differ significantly at adjustedp < .05 using the Holm’s test. 2 5 0 0 2000 1 5 0 0 -I 1000 5 0 0 0 •TAF •ITAF PMS1 PMS2 PMS3 PMS4 Figure 6. Optimality of resource allocation by condition Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 9 H9: When rewarded with SFM alone, TAF managers allocate resources better than ITAF managers. When rewarded with SFM alone (PMS1), TAF managers (M= 431.44, SD = 315.14) allocated resources better than ITAF managers (M = 2302.67, SD = 1153.79), t = -8.35, adjustedp < .01 (one-tailed), Wilcoxon Z = -4.79, p < .01 (one tailed). H9 was supported, corresponding to the results for HI. Additional contrasts were conducted to assess whether the difference in optimality of resource allocation between managers of TAF and ITAF would change under different PMS. When VD information was added (PMS2), when VD weights were added (PMS3), and when VDs were rewarded (PMS4), TAF managers still allocated resources significantly better than ITAF managers. H10 (Learning effect from having VD information): When VD information is provided as compared to when it is not, ITAF managers improve their allocation of resources more than TAF managers do. When VD information was provided as compared to when it was not, both TAF managers (t = -0.19, adjustedp ~ 1.00 [one-tailed], Wilcoxon Z = -0.14,p = .45 [one-tailed]) and ITAF managers (t = -1.38, adjustedp = .43 [one-tailed], Wilcoxon Z - -0.95, p = .17 [one-tailed]) did not allocate resources better. H10 was not supported, corresponding to the results for H2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 H ll (Learning effect from having VD weights): When VD weights are provided as compared to when they are not, ITAF managers improve their allocation of resources more than TAF managers do. When VD weights were provided as compared to when they were not, both TAF managers (t = 0.02, adjusted p = 1.00 [one-tailed], Wilcoxon Z = 0.14, p — ,44[one-tailed]) and ITAF managers (t = 0.01, adjusted p = 1.00 [one-tailed], Wilcoxon Z = 0.22, p = .41 [one-tailed]) did not allocate resources better. H ll was not supported, corresponding to the results for H3. H12 (Effect from VD rewards): When VDs are rewarded as compared to when they are not, ITAF managers improve their allocation of resources more than TAF managers do. When VDs were rewarded as compared to when they were not, TAF managers did not allocate resources more optimally (t = -0 .0 2 , adjusted p = 1.00 [one-tailed], Wilcoxon Z - -0.14, p = .44 [one-tailed]) while ITAF managers (t = - 2.66, adjusted p = .03 [one-tailed], Wilcoxon Z = -1.99, p = .02 [one-tailed]) allocated resources more optimally. H I2 was supported, corresponding to the results for H4. The greater improvement in performance for ITAF managers compared to TAF managers in H4 when VDs were rewarded was not due to greater improvement Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 in learning but was due to better motivation of ITAF managers manifested through improved resource allocation. Additional Analyses Learning as a Covariate To further separate the learning and motivational effects, an ANCOVA was conducted on performance using learning as a covariate to control for the effects of learning on performance (Table 12). The results were similar to the ANOVA on performance with performance being significantly different across conditions (F = 11.77, p < .01) and the Firm X PMS interaction effect being significant (F = 2.89, p = .04). Learning, as a covariate, was not significant, F = 0.28, p = .60. Table 12 ANCOVA for Performance with Learning as a Covariate Source SS df MS F P Model 7552.69 8 944.09 1 1 - 7 7 ** < . 0 1 Error 12353.14 154 80.22 Corrected Total 19905.83 162 Firm 5062.36 1 5062.36 63.11** < . 0 1 PMS 699.13 3 233.04 2.91* .04 Firm X PMS 696.66 3 232.22 2.89* .04 Learning 22.15 1 22.15 0.28 .60 Note. R2 = .38 for Model *p < .05, **p < .01 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 The results for the multiple contrasts of performance adjusted for learning were similar to that for the multiple contrasts of performance unadjusted for learning. Table 13 and Figure 7 displays the results. Table 13 Performance (Adjusted for Learning) by Condition Firm PMS1 PMS2 PMS3 PMS4 TAF 134.19a 132.44a 134.06a 133.98a ITAF 116.94b 120.79b 120.73b 128.17, Note. Values enclosed in parentheses represent standard deviation. Means with different subscripts differ significantly at adjustedp < .05 using the Holm’s test. < a a > Ph T 3 U + - » 3 < •TAF •ITAF 140 135 130 125 120 115 110 PMS1 PMS2 PMS3 PMS4 Figure 7. Performance (adjusted for learning) by condition Overall, the data supported the proposition that the greater improvement in performance for ITAF managers relative to TAF managers when VDs were rewarded was a result of better motivation rather than better learning. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 Prior Beliefs as Covariates ANCOVA was conducted on performance (Table 14) and learning (Table 15), using the prior beliefs of subjects as covariates to control for the effects of prior beliefs on performance and learning. The results were similar to the ANOVA on performance and learning, with performance still being significantly different across conditions (F = 9.49, p < .01), and learning still being significantly different across conditions (F= 3.52,p < .01). Table 14 ANCOVA for Performance with Prior Beliefs as Covariates Source SS 4f MS F P Model 8137.41 1 1 739.76 9 4 9 *** < . 0 1 Error 11768.42 151 77.94 Corrected Total 19905.83 162 Firm 5576.47 1 5576.47 71.55*** < . 0 1 PMS 683.61 3 227.87 2.92** .04 Firm X PMS 570.34 3 190.11 2.44* .07 prc 0.71 1 0.71 0 . 0 1 .92 prf 306.08 1 306.08 3.93** .05 ppc 240.09 1 240.09 3.08* .08 ppf 2.82 1 2.82 0.04 .85 Note. R2= .41 for Model. *p < .10, **p < .05, ***p < .01 prc = size of gross returns of current period R&D investment in the current period prf = size of gross returns of current period R&D investment in future periods ppc = size of gross returns of current period P&E investment in the current period ppf= size of gross returns of current period P&E investment in future periods prc, prf, ppc, and ppf are rated on a 5-point scale (1 = No returns, 5 = Very large returns) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 Table 15 ANCOVA for Learning with Prior Beliefs as Covariates Source SS df MS F P Model 55.17 1 1 5.02 3.52** <.01 Error 215.03 151 1.42 Corrected Total 270.20 162 Firm 36.55 1 36.55 25.67** <.01 PMS 4.99 3 1.66 1.17 .32 Firm X PMS 7.39 3 2.46 1.73 .16 prc 0.24 1 0.24 0.17 .68 prf 0.72 1 0.72 0.50 .48 ppc 2.48 1 2.48 1.74 .19 ppf 0.12 1 0.11 0.08 .78 Note. R2- .20 for Model. *p < .05, **p < .01 prc = size of gross returns of current period R&D investment in the current period prf = size of gross returns of current period R&D investment in future periods ppc = size of gross returns of current period P&E investment in the current period ppf = size of gross returns of current period P&E investment in future periods prc, prf, ppc, and ppf are rated on a 5-point scale (1 = No returns, 5 = Very large returns) The results for the multiple contrasts of performance (Table 16 and Figure 8 ) and learning (Table 17 and Figure 9) adjusted for prior beliefs were similar to the multiple contrasts of performance and learning unadjusted for learning. Hence, the prior beliefs of subjects with respect to the relative importance of the investments and the lagged impact of the investments on firm profits did not explain the performance and learning differences across conditions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 Table 16 Performance (Adjusted for Prior Beliefs) by Condition Firm P M S 1 P M S 2 P M S 3 P M S 4 TAF 1 3 3 .8 4 a 1 3 3 .0 4 a 1 3 3 .7 3 a 134.37a ITAF 1 1 6 .6 6 b 121.36b 1 20.59b 1 2 7 .7 1 , Note. Means with different subscripts differ significantly at adjustedp < .05 using the Holm’s test. 1 4 0 1 3 5 1 3 0 1 2 5 120 1 1 5 110 PMS1 PMS2 PMS3 PMS4 Figure 8. Performance (adjusted for prior beliefs) by condition tS u C L ) Oh < u a s 3 tE P < •TAF ■ITAF Table 17 Learning (Adjusted for Prior Beliefs) by Condition Firm P M S 1 P M S 2 P M S 3 P M S 4 TAF 4 .0 8 a 4.7Q a 3 .9 2 b 3 .7 0 b ITAF 3.20b 2 .9 9 b 3.29b 2 .9 8 b Note. Means with different subscripts differ significantly at adjusted p < .10 using the Holm’s test. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 so S S 3 < D h J - 0 & 0 3 3 tT < 5 i 4.5 4 3.5 3 2.5 •TAF -ITAF P M S 1 P M S 2 P M S 3 P M S 4 Figure 9. Learning (adjusted for prior beliefs) by condition Heuristics Used In Learning The heuristics used by the subjects to understand the relationships between their investments and performance were analyzed. Among the 12 categories of heuristics identified by Hutchinson and Alba (1997), subjects in this study employed the recent versus initial prototypes heuristic (M = 28.34, SD - 21.14), the recent vs. initial prototypes heuristic adjusted for time lag (M - 24.26, SD = 19.91), the adjacent-rows differences heuristic (M = 25.09, SD = 16.51), the best-exemplars heuristic (M = 19.52, SD = 18.11), and the best vs. worst prototypes heuristic (M = 0.09, SD= 1.17). The use of the recent versus initial prototypes heuristic (.F = 0.41, p - 0.90), the recent vs. initial prototypes heuristic adjusted for time lag (F = 0.79, p = 0.60), the adjacent-rows differences heuristic (F = 0.68, p - 0.69), and the best vs. worst prototypes heuristic (F = 1.09, p = 0.37) were not significantly different across Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 conditions. Only the use of the best-exemplars heuristic (F = 2.76, p < .01) was significantly different across conditions with a significant PMS effect (F — 5.67, p < .01). The best-exemplars heuristic was used most in PMS3 (M = 28.72, SD = 26.33), next in PMS2 (M = 18.45, SD = 14.50), next in PMS4 (M = 17.24, £D = 13.09), and the least in PMS1 (M = 13.56, = 11.32). The heuristics utilized by subjects might explain why they did not improve their learning in the study. In general, subjects used chunk-based heuristics (recent vs. initial prototypes heuristics, recent vs. initial prototypes heuristics adjusted for time lag, and best vs. worst prototypes heuristics) more often than difference-based heuristics (adjacent-row differences heuristic) and exemplar-based heuristics (best- exemplars heuristic). Also, in sampling observations, subjects used a locational criteria based on the location of the observation more than an informational criteria based on the value of the observation. Subjects tried to detect trends in their investments and performance across time periods by grouping observations and forming prototypes of observations from initial periods and later periods. The use of such chunk-based representations was logical given that subjects were probably trying to simplify information processing by grouping observations. While some subjects factor in a time lag when using chunk-based representations, other subjects did not take into account investments affecting firm profits only three periods later. Merely observing the trend of investments and performance across time periods without accounting for the lag between investments and performance would impede learning. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 8 Subjects also varied their investments in the current period from the prior period and observed the change in performance in the current period from the prior period. The use of such difference-based representation would be effective only when performance did not lag investments. Comparing performance changes with the corresponding investment changes from one period to the next period would hinder the detection of the lagged relationships between investments and firm profits. Power Analyses H2 - H3 for Performance, H6 - H 8 for Learning, and H10 - H ll for Optimality of Resource Allocation were not supported. I conducted a power analyses to assess whether the failure to detect significant effects was due to insufficient sample sizes that resulted in a lack of power. The effect sizes, d, and the corresponding power to detect effects at a = .05, 1 - (5 , of the contrasts tested are reported in Table 18. Power ranged from to .05 to 1.00 for detecting differences in Performance and Optimality of Resource Allocation, and from .11 to 1.00 for detecting differences in Learning. If there were a systematic lack of power across all contrasts, the insignificant effects could have been a result of insufficient sample sizes. However, there was no systematic lack of power across all contrasts. If there were insignificant effects across all contrasts, it would have suggested that the sample sizes were too small to detect any effects. However, significant effects were detected for some contrasts. The power analyses suggested that the insignificant contrasts were primarily due to small effect sizes rather than insufficient sample sizes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Table 18 Effect Sizes and Power of Contrasts Panel A: Performance Contrast d 1 - P Condition 1 vs. Condition 5 2 . 0 1 1 . 0 0 Condition 2 vs. Condition 6 1.40 .96 Condition 3 vs. Condition 7 1.55 1 . 0 0 Condition 4 vs. Condition 8 0.69 1 . 0 0 Condition 1 vs. Condition 2 0.18 .29 Condition 5 vs. Condition 6 0.44 . 2 0 Condition 2 vs. Condition 3 0.15 .25 Condition 6 vs. Condition 7 0.003 .05 Condition 3 vs. Condition 4 0 . 0 2 .07 Condition 7 vs. Condition 8 0.84 .78 Panel B: Learning Contrast d 1 - P Condition 1 vs. Condition 5 0.74 .72 Condition 2 vs. Condition 6 1.43 1 . 0 0 Condition 3 vs. Condition 7 0.45 .43 Condition 4 vs. Condition 8 0.58 .46 Condition 1 vs. Condition 2 0.57 .51 Condition 5 vs. Condition 6 0 . 1 2 . 1 1 Condition 2 vs. Condition 3 0.74 .72 Condition 6 vs. Condition 7 0.24 . 2 2 Condition 3 vs. Condition 4 0.16 . 1 2 Condition 7 vs. Condition 8 0.29 . 2 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Panel C: Optimality of Resource Allocation 110 Contrast d 1-P Condition 1 vs. Condition 5 2.71 1 . 0 0 Condition 2 vs. Condition 6 2.32 1 . 0 0 Condition 3 vs. Condition 7 2.32 1 . 0 0 Condition 4 vs. Condition 8 1.48 1 . 0 0 Condition 1 vs. Condition 2 0.06 . 1 1 Condition 5 vs. Condition 6 0.45 . 2 1 Condition 2 vs. Condition 3 0 . 0 1 .05 Condition 6 vs. Condition 7 0.003 .05 Condition 3 vs. Condition 4 0 . 0 1 .05 Condition 7 vs. Condition 8 0.84 .79 Note. Small, medium, and large effect sizes are defined by Cohen (1988) as d = .20, .50, and .80, respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill CHAPTER 5: CONCLUSIONS Summary of Results The results in this study indicate that adding VDs to a SFM improves managerial performance and reduces performance variability under two specific conditions. First, VDs improve performance only when they are explicitly rewarded in the compensation scheme. Merely, providing information on the performance of VDs or information on the relative importance of VDs does not lead to better performance. Second, when VDs are rewarded, performance improves only for managers of firms that depend more on intangible assets but not for managers of firms relying more on tangible assets. The results for both the learning variable and the performance variable suggest that the performance improvement is a result of improved motivation when VDs are rewarded and not of improved learning when VD information or VD weights are provided or when VDs are rewarded. When VD information or VD weights are given to managers, they do not improve their learning or performance. When VDs are rewarded, the learning of all managers does not improve but the performance of managers in intangible assets firms improves. The ANCOVA analyses also indicate that learning does not explain the performance differences across conditions. The ANCOVA analyses further show that the prior beliefs of managers regarding the relative importance of different types of investments in the current period and future periods do not explain the results obtained in this study. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 The results for the attention of managers may explain why managers do not learn better when VD information or VD weights are provided and when VDs are rewarded. In general, adding VDs to a PMS does not seem to induce managers to pay greater attention to the long-term effects of their investments and the effects of the more important investment on performance in their firm. The heuristics used by managers may also explain why they do not learn when VDs are added to a SFM. To simplify information processing, managers tend to use chunk-based and difference-based heuristics to understand the relationships between their investments and firm profits. Moreover, when presented with data, managers tend to sample observations based on the spatial characteristics of the variables rather than the values of the variables. These heuristics make it more difficult for managers to detect the lagged relationships between investments and firm profits and impede their learning. Limitations of Study Inference o f Learning from Improvement in Performance My study delineates the performance effects that result from different feedback versus different incentives and infers learning from any improvement in performance under different feedback. However, my study does not test the permanence of performance effects. Since learning is the acquisition of underlying knowledge and skills, the performance change that results from learning is a relatively permanent and stable change (Christina & Bjork, 1991; Goodman, 1998; Salmoni et al., 1984; Schmidt & Bjork, 1992). This is contrasted with temporary Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 performance improvement that may disappear over time or under different conditions. Studies have been designed to try to differentiate between transient and permanent performance changes (e.g., Goodman, 1998; Schmidt, 1989). Typically, feedback that is provided during the practice session is removed at a later time to assess the stability of the performance change. Another way to identify permanent learning effects is to test for the transfer of skills to a similar task. In my study, I did not remove the feedback and test the stability of the performance change or test the transfer of skills using another task. I did not want to overburden subjects with additional tasks. Moreover, in the context of managerial performance in an organization, feedback is constantly provided and there are rarely situations where feedback is not given on performance. Performance measurement and evaluation are embedded processes in the organization. Even when the task or conditions change, feedback is still needed to help managers understand what has changed and adjust their strategies as appropriate. Managerial tasks are different from some tasks where even if external feedback about results are not provided, people would still have access to other forms of feedback to assess the appropriateness of their response. For example, in physical tasks, intrinsic feedback could naturally arise when a physical movement is made (Salmoni et al., 1984; Schmidt, 1991). Hence, people playing the piano may hear their own playing and assess whether they are playing well rather than depend on a listener to inform them of the quality of their playing. In the case of managers, they have to depend on feedback generated from the PMS to assess their Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 performance. Firms do not provide feedback with the intention of helping managers learn and then remove the feedback after the manager has learnt. Limitations o f the Experiment Setting While I try to create an experimental setting that is as realistic as possible in this study, the nature of an experiment still entails a setting that is stylized and abstract relative to the real world. As such, generalization of the findings in this study should be done with caution. For example, in the experimental setting, subjects have no other sources of information to assess the relationships between investments and firm profits other than the performance measures in the PMS. In the real world, managers have other sources of information such as industry information, competitor information, knowledge of co-workers, and the established firm strategy. The learning of managers in the real world may be enhanced by these other sources of information. On the other hand, in the real world, the performance of the firm is affected by many other exogenous variables such as economic factors, actions of competitors, and actions of other divisions, over which the manager has little control. In such situations, the learning of managers may be impeded. However, limiting the available sources of information to the performance measures in the PMS and ensuring that only the manager’s actions affect firm performance enable the isolation of the sources of performance differences and is one of the benefits of a controlled experiment. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 Subjects also bring to the experiment preconceived ideas about the relative importance of different types of investments and the lagged impact of investments on performance in the semi-conductor industry. This is controlled by the random assignment of subjects to conditions. Moreover, the ANCOVA analyses using the prior beliefs of subjects as covariates do not change the results. Future Research Directions The use of nonfmancial measures in PMS is a field where relatively less research has been undertaken. In particular, the impact of nonfmancial measures on the individual manager has not been examined extensively. Much more can be done to understand the consequences of using nonfmancial measures and the different conditions under which nonfmancial measures may be more effective. The following are some specific research ideas that are extensions of the current study. This study only examines the effects of different PMSs between subjects. An extension of this study is to examine the effects of changing PMS and assess performance changes within subject. Managers currently working under one PMS and preparing to change to another PMS may experience different learning and motivational effects from managers who have not had prior experiences with other PMSs. Changing PMS can take several forms. For example, a PMS can be changed from one in which only financial measures are rewarded to one in which nonfmancial measures are rewarded as well. A PMS can also be changed when either the performance measures or the weights on the measures are changed when firms alter or fine-tune their strategy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 In this study, subjects are given the accurate weights to be placed on each VD. However, when firms introduce nonfmancial measures into the PMS, they are usually not certain of the appropriate weights to be placed on different performance measures in the PMS. Another extension of this study involves testing the effects of inaccurate weights on the VDs in the PMS. Future research may examine whether inaccurate weights on the VDs adversely affects the learning and performance of managers. This study examines a performance evaluation setting where there are no strategic interactions between the PMS and the manager. The PMS is constant throughout the experiment and does not change regardless of the investment choices and the performance of the manager. In reality, the PMS is not static and may change strategically depending on the manager’s actions and performance, with the intention of improving outcomes for the firm. A manager working under a strategic PMS would act differently compared to a manager working under a static and non- strategic PMS. Hence, it will be interesting to examine the strategic interplay between the PMS and the manager using a game theoretic framework. Finally, this study highlights the lack of learning in managers when VDs are added to a PMS and provides some preliminary evidence about the type of heuristics that managers use when they are trying to understand the relationships between performance measures. Future research may examine how the PMS can be designed to enhance the learning and performance of managers by taking into consideration the heuristics that managers employ. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 BIBLIOGRAPHY Abemethy, M. A., & Lillis, A. M. (1995). The impact of manufacturing flexibility on management control system design. Accounting, Organizations and Society, 20, 41-258. Abowd, J. (1990). Does performance-based managerial compensation affect corporate performance? Industrial and Labor Relations Review, 43, 52-73. Adelson, B., & Soloway, E. (1988). A model of software design. In M. Chi, R. Glaser & M. Farr (Eds.), The nature o f expertise (pp. 185-205). Hillsdale, NJ: Erlbaum. American Institute of Certified Public Accountants (AICPA), Special Committee on Financial Reporting. (1994). Improving business reporting - A customer focus: Meeting the information needs o f investors and creditors. New York, NY: AICPA. American Institute of Certified Public Accountants (AICPA), & Maisel, L. S. (2001). Performance measurement practices survey results. New York, NY: AICPA. Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369-460. Anderson, E. W., Fomell, C., & Lehmann, D. R. (1994). Customer satisfaction, market share, and profitability: Findings from Sweden. Journal o f Marketing Research, 58, 53-66. Anderson, J. R., Greeno, J. G., Kline, P. J., & Neves, D. M. (1981). Acquisition of problem-solving skill. In J. R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 191-230). Hillsdale, NJ: Erlbaum. Anonymous. (1981, September 21). Nucor’s big-buck incentives. Business Week (Industrial/technology edition), 2706, 42Z. Anzai, Y., & Simon, H. A. (1979). The theory of learning by doing. Psychological Review, 86, 124-140. Atkinson, J. W. (1964). An introduction to motivation. Princeton, NJ: Van Nostrand. Balzer, W. K., Doherty, M. E., & O’ Connor, R., Jr. (1989). Effects of cognitive feedback on performance. Psychological Bulletin, 106, 410-433. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Balzer, W. K., Sulsky, L. M., Hammer, L. B., & Sumner, K. R. (1992). Task information, cognitive information, or functional validity information: Which components of cognitive feedback affect performance? Organizational Behavior and Human Decision Processes, 53, 35-54. Banker, R. D., & Datar, S. (1989). Sensitivity, precision, and linear aggregation of signals for performance evaluation. Journal o f Accounting Research, 27, 21- 37. Banker, R. D., Lee, S., & Potter, G. (1996). A field study of the impact of a performance-based incentive plan. Journal o f Accounting & Economics, 21, 21-39. Banker, R. D., Potter, G., & Schroeder, R. G. (1993). Manufacturing performance reporting for continuous quality improvement. Management International Review, 33(Special Issue), 69-85. Banker, R. D., Potter, G., & Srinivasan, D. (2000). An empirical investigation of an incentive plan that includes nonfinancial performance measures. The Accounting Review, 75, 65-92. Becker, L. J. (1978). Joint effect of feedback and goal setting on performance: A field study of residential energy conservation. Journal o f Applied Psychology, 63, 428-433. Bedard, J., & Biggs, S. (1991). Pattern recognition, hypotheses generation, and auditor performance in an analytical task. The Accounting Review, 66, 622- 642. Behn, B. K., & Riley, R. A., Jr. (1999). Using nonfinancial information to predict financial performance: The case of the U.S. airline industry. Journal o f Accounting, Auditing & Finance, 14, 29-56. Benzion, U., Rapoport, A., & Yagil, J. (1989). Discount rates inferred from decisions: An experimental study. Management Science, 35, 270-284. Berry, D. C., & Broadbent, D. E. (1988). Interactive tasks and the implicit-explicit distinction. British Journal o f Psychology, 79, 251-272. Bimbaum, M. H. (1999). Testing critical properties of decision making on the Internet. Psychological Science, 10, 399-407. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Bimbaum, M. H. (2000). Decision making in the lab and on the Web. In M. H. Bimbaum (Ed.), Psychological experiments on the Internet (pp. 3-34). San Diego, CA: Academic Press. Bonner, S. E. (1994). A model of audit task complexity. Accounting, Organizations and Society, 19, 213-234. Bonner, S. E., Hastie, R., Sprinkle, G. B., & Young, M. (2000). A review of the effects of financial incentives on performance in laboratory tasks: Implications for management accounting. Journal o f Management Accounting Research, 12, 19-64. Bonner, S. E., & Lewis, B. L. (1990). Determinants of auditor expertise. Journal o f Accounting Research, 2§(Suppl.), 1-20. Bonner, S. E., & Walker, P. L. (1994). The effects of instruction and experience on the acquisition of auditing knowledge. The Accounting Review, 69, 157-178. Brehmer, B. (1974). Hypotheses about relations between scaled variables in the learning of probabilistic inference tasks. Organizational Behavior and Human Performance, 11, 1-27. Brehmer, B. (1979). Preliminaries to a psychology of inference. Scandinavian Journal o f Psychology, 20, 193-210. Brehmer, B. (1990). Strategies in real-time, dynamic decision making. In R. Hogarth (Ed.), Insights in decision making: A tribute to Hillel J. Einhorn (pp. 262- 279). Chicago, IL: University of Chicago Press. Brehmer, B. (1992). Dynamic decision making: Human control of complex systems. Acta Psychologica, 81, 211-241. Brehmer, B. (1995). Feedback delays in complex dynamic decision tasks. In P. Frensch & J. Funke (Eds.), Complex problem solving: The European perspective (pp. 103-130). Hillsdale, NJ: Erlbaum. Brehmer, B., & Allard, R. (1991). Dynamic decision making: The effects of task complexity and feedback delay. In J. Rasmussen, B. Brehmer & J. Leplat (Eds.), Distributed decision making: Cognitive models fo r cooperative work (pp. 319-333). New York, NY: John Wiley & Sons. Brehmer, B., & Qvamstrom, G. (1976). Information integration and subjective weights in multiple-cue judgements. Organizational Behavior and Human Performance, 17, 118-126. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 Broniarczyk, S. M., & Alba, J. W. (1994). Theory versus data in prediction and correlation tasks. Organizational Behavior and Human Decision Processes, 57, 117-139. Brunswick, E. (1955). Representative design and probabilistic theory. Psychological Review, 62, 236-242. Brunswick, E. (1956). Perception and the representative design o f psychological experiments. Berkeley, CA: University of California Press. Buchanan, T. (2000). Potential of the Internet for personality research. In M. H Bimbaum (Ed.), Psychological experiments on the Internet (pp. 121-140). San Diego, CA: Academic Press. Buchanan, T., & Smith, J. L. (1999). Using the Internet for psychological research: Personality testing on the World-Wide Web. British Journal o f Psychology, 90, 125-144. Bushee, B. J. (1998). The influence of institutional investors on myopic R&D investment behavior. The Accounting Review, 73, 305-333. Bushman, R. M., Indjejikian, R. J., & Smith, A. (1996). CEO compensation: The role of individual performance evaluation. Journal o f Accounting & Economics, 21,161-193. Carol, K. (1986). Performance pay: Cash on delivery. Director, 40(4), 34-37. Chase, W. G., & Simon, H. A. (1973). The mind's eye in chess. In W.G. Chase (Ed.), Visual information processing (pp. 215-281). New York, NY: Academic Press. Chi, M. T. H., Glaser, R., & Rees, E. (1982). Expertise in problem solving. In R. J. Sternberg (Ed.), Advances in the Psychology o f Human Intelligence: Vol. 1 (pp. 7-75). Hillsdale, NJ: Erlbaum. Chidester, T. R., & Grigsby, W. C. (1984). A meta-analysis of the goal setting performance literature. Academy o f Management Proceedings, 202-206. Christina, R. W., & Bjork, R. A. (1991). Optimizing long-term retention and transfer. In D. Druckman & R. A. Bjork (Eds.), In the mind’ s eye: Enhancing human performance (pp. 23-55). Washington, DC: National Academy Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121 Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Conover, W. J. (1980). Practical nonparametric statistics (2nd ed.). New York, NY: John Wiley & Sons. Cook, F. W. (1987). Reintroducing risk into executive pay. Directors and Boards, 11(4), 27-28. Deane, D. H., Hammond, K. R., & Summers, D. A. (1972). Acquisition and application of knowledge in complex inference tasks. Journal of Experimental Psychology, 92, 20-26. Dechow, P., & Sloan, R. (1991). Executive incentives and the horizon problem. Journal o f Accounting & Economics, 14, 51-89. Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York, NY: Plenum Press. DeNisi, A. S., & Williams, K. J. (1988). Cognitive approaches to performance appraisal. In G. Ferris & K. Rowland (Eds.), Research in personnel and human resource management (pp. 109-155). Greenwich, CT: JAI Press. Diehl, E., & Sterman, J. D. (1995). Effects of feedback complexity on dynamic decision making. Organizational Behavior and Human Decision Processes, 6 2 ,198-215. Eccles, R. G. (1991). The performance measurement manifesto. Harvard Business Review, 69(2), 131-137. Edwards, W. (1962). Dynamic decision theory and probabilistic information processing. Human Factors, 4, 59-73. Ehrbar, A. (1998). EVA: The real key to creating wealth. New York, NY: John Wiley & Sons. Einhom, H., & Hogarth, R. (1986). Judging probable cause. Psychological Bulletin, 99, 3-19. Eisenberger, R. & Cameron, J. (1996). Detrimental effects of reward: Reality or myth? American Psychologist, 51, 1153-1166. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122 Eisenberg, R. & Cameron, J. (1998). Reward, intrinsic interest, and creativity: New findings. American Psychologist, 53, 676-679. Eisenberger, R., Rhoades, L., & Cameron, J. (1999). Does pay for performance increase or decrease perceived self-determination and intrinsic motivation? Journal o f Personality and Social Psychology, 77, 1026-1040. Erez, M. (1977). Feedback: A necessary condition for goal setting-performance relationship. Journal o f Applied Psychology, 62, 624-627. Feltham, G. A., & Xie, J. (1994). Performance measure congruity and diversity in multi-task principal/agent relations. The Accounting Review, 69, 429-453. Gibson, F. P., Fichman, M., & Plaut, D. C. (1997). Learning in dynamic decision tasks: Computational model and empirical evidence. Organizational Behavior and Human Decision Processes, 71, 1-35. Goodman, J. S. (1998). The interactive effects of task and external feedback on practice performance and learning. Organizational Behavior and Human Decision Processes, 76, 223-252. Govindarajan, V., & Gupta, A. (1985). Linking control systems to business unit strategy: Impact on performance. Accounting, Organizations and Society, 10, 51-66. Grinblatt, M., & Titman, S. (1987). How clients can win the gaming game. Journal o f Portfolio Management. 13(4), 14-20. Groen, G., & Patel, V. (1988) The relationship between comprehension and reasoning in medical expertise. In M. Chi, R. Glaser & M. Farr (Eds.), The nature o f expertise (pp. 287-310). Hillsdale, NJ: Erlbaum. Guidry, F., Leone, A. J., & Rock, S. (1999). Eamings-based bonus plans and earnings management by business-unit managers. Journal o f Accounting & Economics, 26, 113-142. Hammond, K. R., Hursch, C. J., & Todd, F. J. (1964). Analyzing the components of clinical inference. Psychological Review, 71, 438-456. Hammond, K. R., & Summers, D. A. (1965). Cognitive dependence on linear and nonlinear cues. Psychological Review, 72, 215-224. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 Hammond, K. R., Summers, D. A., & Deane, D. H. (1973). Negative effects of outcome-feedback in multiple-cue probability learning. Organizational Behavior and Human Performance, 9, 30-34. Hayes, N. A., & Broadbent, E. E. (1988). Two modes of learning for interactive tasks. Cognition, 28, 249-276. Healy, P. (1985). The effect of bonus schemes on accounting decisions. Journal o f Accounting & Economics, 7, 85-107. Hewson, C. M., Laurent, D., & Vogel, C. M. (1996). Proper methodologies for psychological and sociological studies conducted via the Internet. Behavior Research Methods, Instruments, & Computers, 28(2), 186-191. Hoffman, P. J., Earle, T. C., & Slovic, P. (1981). Multidimensional functional learning (MFL) and some new conceptions of feedback. Organizational Behavior and Human Performance, 27, 75-102. Hogarth, R. M. (1981). Beyond discrete biases: Functional and dysfunctional aspects of judgment heuristics. Psychological Bulletin, 90, 197-217. Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal o f Statistics, 6, 65-70. Holmstrom, B. (1979). Moral hazard and observability. Bell Journal o f Economics, 10, 74-91. Holmstrom, B. & Milgrom, P. (1991). Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design. Journal o f Law, Economics, & Organization, 7, 24-52. Holthausen, R., Larcker, D., & Sloan, R. (1995). Annual bonus schemes and the manipulation of earnings. Journal o f Accounting & Economics, 19, 29-74. Hopwood, A. G. (1972). An empirical study of the role of accounting data in performance evaluation. Journal o f Accounting Research, iO(Suppl.), 156- 182. Huber, O. (1995). Complex problem solving as multistage decision making. In P.A. Frensch & J. Funke (Eds.), Complex problem solving: The European perspective (pp. 151-173). Hillsdale, NJ: Erlbaum. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 Hursch, C. J., Hammond, K. R., & Hursch, J. L. (1964). Some methodological considerations in multiple-cue probability studies. Psychological Review, 71, 42-60. Hutchinson, J. W., & Alba, J. W. (1997). Heuristics and biases in the “eyeballing” of data: The effects of context on intuitive correlation assessment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 23, 591-621. Indjejikian, R. J. (1999). Performance evaluation and compensation research: An agency perspective. Accounting Horizons, 13, 147-157. Ittner, C. D., & Larkcer, D. F. (1995). Total quality management and the choice of information and reward systems. Journal o f Accounting Research, i5(Suppl.), 1-34. Ittner, C. D., & Larcker, D. F. (1997). Quality strategy, strategic control systems, and organizational performance. Accounting, Organizations and Society, 22, 293- 314. Ittner, C. D., & Larcker, D. F. (1998a). Innovations in performance measurement: Trends and research implications. Journal o f Management Accounting Research, 10, 205-238. Ittner, C. D., & Larcker, D. F. (1998b). Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction. Journal o f Accounting Research, 3<5(Suppl.), 1-35. Ittner, C. D., Larcker, D. F., & Raj an, M. V. (1997). The choice of performance measures in annual bonus contracts. The Accounting Review, 72, 231-255. Johnson, H. T., & Kaplan, R. S. (1987). Relevance lost: The rise and fall o f management accounting. Boston, MA: Harvard Business School Press Kahneman, D., & Tver sky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-291. Kaplan, R. S., & Norton, D. P. (1992). The balanced scorecard - measures that drive performance. Harvard Business Review, 70(1), 71-79. Kaplan, R. S., & Norton, D. P. (1993). Putting the balanced scorecard to work. Harvard Business Review, 71(5), 134-147. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 Kaplan, R. S., & Norton, D. P. (1996). Using the balanced scorecard as a strategic management system. Harvard Business Review, 74(1), 75-85. Kaplan, R. S., & Norton, D. P. (2001a). The strategy focused organization: How balanced scorecard companies thrive in the new business environment. Boston, MA: Harvard Business School Press. Kaplan, R. S., & Norton, D. P. (2001b). Transforming the balanced scorecard from performance measurement to strategic management: Part I. Accounting Horizons, 15, 87-104. Kaplan, R. S., & Norton, D. P. (2001c). Transforming the balanced scorecard from performance measurement to strategic management: Part II. Accounting Horizons, 15, 147-160. Kirk, R. E. (1995). Multiple Comparisons Test. In Experimental design: Procedures fo r the behavioral sciences (pp. 113-163). New York, NY: Brooks/Cole. Kleinmuntz, D. N. (1993). Information processing and misperceptions of the implications of feedback in dynamic decision making. System Dynamics Review, 9, 223-237. Kolb, D. A., & Boyatzis, R. E. (1970). Goal-setting and self-directed behavior change. Human Relations, 23, 439-457. Komaki, J., Barwick, K. D., & Scott, L. R. (1978). A behavioral approach to occupational safety: Pinpointing and reinforcing safe performance in a food manufacturing plant. Journal o f Applied Psychology, 63, 434-445. Krantz, J. H. (2002). Psychological research on the net. Retrieved December 3, 2002, from http://psych.hanover.edu/research/exponnet.html Krantz, J. H., Ballard, J., & Scher, J. (1997). Comparing the results of laboratory and World-Wide Web samples on the determinants of female attractiveness. Behavior Research Methods, Instruments, & Computers, 29, 264-269. Krantz, J. H., & Dalai, R. (2000). Validity of Web-based psychological research. In M. H. Bimbaum (Ed.), Psychological Experiments on the Internet (pp. 35- 60). San Diego, CA: Academic Press. Krippendorff, K. (1980). Content analysis: An introduction to its methodology. Beverly Hills, CA: Sage. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 Langley, P., & Simon, H. A. (1981). The central role of learning in cognition. In J. R. Anderson (Ed.),. Cognitive skills and their acquisition (pp. 361-380). Hillsdale, NJ: Erlbaum. Larcker, D. (1983). The association between performance plan adoption and corporate investment. Journal o f Accounting & Economics, 5, 3-30. Larkin, J. H. (1981). Enriching formal knowledge: A model for learning to solve textbook physics problems. In J. R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 311 -334). Hillsdale, NJ: Erlbaum. Lee, J. W., & Yates, J. F. (1992). How quantity judgement changes as the number of cues increases: An analytical framework and review. Psychological Bulletin, 112, 363-377. Lev, B. (1989). On the usefulness of earnings and earnings research: Lessons and directions from two decades of empirical research. Journal o f Accounting Research, 27(Suppl.), 153-192. Lev, B. (2000). Knowledge management: Fad or need? Research Technology Management, 43(5), 9 - 10. Lev, B. (2001). Intangibles: Management, measurement, and reporting. Washington, DC: Brookings Institution Press. Lev, B., & Zarowin, P. (1999). The boundaries of financial reporting and how to extend them. Journal o f Accounting Research, 37, 353-385. Levene, H. (I960). Robust tests for the equality of variances. In I. Olkin (Ed.), Contributions to probability & statistics (pp. 278-292). Palo Alto, CA: Stanford University Press. Lewis, M. W., & Anderson, J. R. (1985). Discrimination of operator schemata in problem solving: Learning from examples. Cognitive Psychology, 17, 26-65. Libby, R., & Luft, J. (1993). Determinants of judgment performance in accounting settings: Ability, knowledge, motivation, and environment. Accounting, Organizations and Society, 18, 425-450. Libby, R., & Tan, H. T. (1994). Modeling the determinants of audit expertise. Accounting, Organizations and Society, 19, 701-716. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 Lindberg, L. A., & Brehmer, B. (1977a). Subjects ’ selection o f feedback information in an inductive inference task (Umea Psychological Rep. No. 122). Sweden: University of Umea, Department of Psychology. Lindberg, L. A., & Brehmer, B. (1977b). Effects o f task information and active feedback control in inductive inference (Umea Psychological Rep No. 123). Sweden: University of Umea, Department of Psychology. Locke, E. A., & Latham, G. P. (1990). A theory o f goal setting and task performance. Englewood Cliffs, NJ: Prentice-Hall. Locke, E. A., Shaw, K., Saari, L., & Latham, G. (1981). Goal-setting and task performance: 1969-1980. Psychological Bulletin, 90, 125-152. Luft, J. L., & Shields, M. D. (2001). Why does fixation persist? Experimental evidence on judgment performance effects of expensing intangibles. The Accounting Review, 76, 561-588. Luft, J. L., & Shields, M. D. (2002). The effects of financial and nonfinancial performance measures on judgment and decision performance. Working Paper. East Lansing: Michigan State University. McKinnon, S. M., & Bums, W. J. (1992). The information mosaic. Boston, MA: Harvard Business School Press. Mento, A. J., Steel, R. P., & Karren, R. J. (1987). A meta-analytic study of the effects of goal setting on task performance: 1964-1984. Organizational Behavior and Human Decision Processes, 39, 52-83. Merchant, K. (1985). Organizational controls and discretionary program decision making: A field study. Accounting, Organizations and Society, 10, 67-85. Merchant, K. (1989). Rewarding Results: Motivating Profit Center Managers. Boston, MA: Harvard Business School Press. Merchant, K. (1990). The effects of financial controls on data manipulation and management myopia. Accounting, Organizations and Society, 15, 297-313. Merchant, K., & Wu, A. (1995). Teco Electric & Machinery Co. Ltd. University of Southern California, School of Business Administration. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 Muchinsky, P. M., & Dudycha, A. L. (1975). Human inference behavior in abstract and meaningful environment. Organizational Behavior and Human Performance, 13, 377-391. Nemeroff, W. F., & Cosentino, J. (1979). Utilizing feedback and goal setting to increase performance appraisal interviewer skills of managers. Academy o f Management Journal, 22, 566-576. Paich, M., & Sterman, J. D. (1993). Boom, bust, and failures to learn in experimental markets. Management Science, 39, 1439-1458. Pasveer, K. A., & Ellard, J. H. (1998). The making of a personality inventory: Help from the WWW. Behavior Research Methods, Instruments, & Computers, 30, 309-313. Perera, S., Harrison, G., & Poole, M. (1997). Customer-focused manufacturing strategy and the use of operations-based non-financial performance measures: A research note. Accounting, Organizations and Society, 22, 557-572. Peterson, D. K., & Pitz, G. F. (1986). Effects of amount of information on predictions of uncertain quantities. Acta Psychologica, 61, 229-241. Pettit, F. A. (1999). Exploring the use of the World Wide Web as a psychology data collection tool. Computers in Human Behavior, 15, 67-71. Reips, U. D. (2000). The Web experiment method: Advantages, disadvantages, and solutions. In M. H. Bimbaum (Ed.), Psychological experiments on the Internet (pp. 89-117). San Diego, CA: Academic Press. Remus, W., O’Connor, M., & Griggs, K. (1996). Does feedback improve the accuracy of recurrent judgmental forecasts? Organizational Behavior and Human Decision Processes, 66, 22-30. Rice, J. A. (1995). Mathematical statistics and data analysis. Belmont, CA: Duxbury Press. Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of results and motor learning: A review and critical reappraisal. Psychological Bulletin, 95, 355-386. Schmidt, K. H., Kleinbeck, U., & Brockmann, W. (1984). Motivational control of motor performance by goal setting in a dual-task situation. Psychological Research, 46, 129-141. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Schmidt, R. A. (1991). Frequent augmented feedback can degrade learning: Evidence and interpretations. In J. Requin & G. E. Steimach (Eds.), Tutorials in motor neurosicence (pp. 59-75). Dordrecht, The Netherlands: Kluwer Academic. Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3, 207-217. Schmidt, R. A., Young, D. E., Swinnen, S., & Shapiro, D. C. (1989). Summary knowledge of results for skill acquisition: Support for the guidance hypothesis. Journal o f Experimental Psychology: Learning, Memory, and Cognition, 15, 352-259. Schmidt, W. C. (1997). World-Wide Web survey result: Benefits, potential problems, and solutions. Behavior Research Methods, Instruments, & Computers, 30, 309-313. Seidler, L. J. (1986). The fundamental limitations of accounting. Journal o f Accounting, Auditing, & Finance, 1, 353-361. Semiconductor Industry Association, (n.d.) Industry Facts & Figures. Retrieved December 27, 2002, from http://www.semichips.org/pre facts.cfm Shanks, D. R., Pearson, S. M., & Dickinson, A. (1989). Temporal contiguity and the judgment of causality by human subjects. The Quarterly Journal o f Experimental Psychology: Section B, 41, 139-159. Shields, M. D. (1983). Effects of information supply and demand on judgment accuracy: Evidence from corporate managers. The Accounting Review, 58, 284-303. Sieling, M. S. (1988). Semiconductor productivity gains linked to multiple innovations. Monthly Labor Review, 111(4), 27-33. Simon, H. A. (1982). Models of Bounded Rationality. Cambridge, MA: MIT Press. Simon, R. (2000). Performance Measurement & Control Systems for Implementing Strategy. Upper Saddle River, NJ: Prentice Hall. Slovic, P., & Lichtenstein, S. (1971). Comparison of Bayesian and Regression Approaches to the study of information processing in judgment. Organizational Behavior and Human Performance, 6, 649-741. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 Smith, T. W. (2002). Standard & Poor’ s Semiconductors Industry Survey: July 18, 2002. New York, NY: Standard & Poor’s. Sniezek, J. A. (1986). The role of variable labels in cue probability learning tasks. Organizational Behavior and Human Decision Processes, 38, 141-161. Stanton, J. M. (1998). An empirical assessment of data collection using the Internet. Personnel Psychology, 51, 709-725. Sterman, J. D. (1989a). Misperceptions of feedback in dynamic decision making. Organizational Behavior and Human Decision Processes, 43, 301-335. Sterman, J. D. (1989b). Modeling managerial behavior: Misperceptions of feedback in a dynamic decision making experiment. Management Science, 35, 321- 339. Sterman, J. D. (1994). Learning in and about complex systems. System Dynamics Review, 10, 291-330. Stewart, G. B. (1991). The quest for value: A guide for senior managers. New York, NY: HarperBusiness. Stocks, M. H., & Harrell, A. (1995). The impact of an increase in accounting information level on the judgment quality of individuals and groups. Accounting, Organizations, and Society, 20, 685-700. Strang, H. R., Lawerence, E.C., & Fowler, P. C. (1978). Effects of assigned goal level and knowledge of results on arithmetic computation: A laboratory study. Journal o f Applied Psychology, 63, 446-450. Summers, S. A., & Hammond, K. R. (1966). Inference behavior in multiple-cue tasks involving both linear and nonlinear relations. Journal o f Experimental Psychology, 71, 751-757. Summers, S. A., Summers, R. C., & Karkau, V. T. (1969). Judgments based on different functional relationships between interacting cues and a criterion. American Journal o f Psychology, 82, 203-211 Suzuki, K. (1985). Knowledge capital and the private rate of return to R&D in Japanese Manufacturing Industries. International Journal o f Industrial Organization, 3, 293-305. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Thaler, R. (1981). Some empirical evidence on dynamic inconsistency. Economic Letters, 8, 201-207. The Hackett Group. (1998). The Book o f Numbers: Planning and Performance Measurement. [Brochure]. Hudson, OH: The Hackett Group. Todd, F. J., & Hammond, K. R. (1965). Differential feedback in a two multiple-cue probability learning tasks. Behavioral Science, 10, 429-435. Tortoriello, R. S. (2002). Standard & Poor’ s Semiconductor Equipment Industry Survey: August 15, 2002. New York: Standard & Poor’s. Tubbs, M. E. (1986). Goal-setting: A meta-analytic examination of the empirical evidence. Journal o f Applied Psychology, 71, 474-483. Tucker, L. R. (1964). A suggested alternative formulation in the developments by Hursch, Hammond, and Hursch, and by Hammond, Hursch, and Todd. Psychological Review, 71, 528-530. Tversky, A., & Kahneman, D. (1982). Causal schemes in judgments under uncertainty. In D. Kahneman, P. Slovic & A. Tversky (Eds.) Judgment under uncertainty: Heuristics and Biases (pp. 117-128). Cambridge, United Kingdom: Cambridge University Press. Varian, Hal. R. (1996). Intermediate microeconomics: A modern approach (4th ed.). New York, NY: Norton. Vroom, V. (1964). Work and Motivation. New York, NY: Wiley. Waller, W. S., & Bishop, R. A. (1990). An experimental study of incentive pay schemes, communication, and intrafirm resource allocation. The Accounting Review, 65, 812-836. Wasserman, E. A., & Neunaber, D. J. (1986). College students’ responding to and rating of contingency relations: The role of temporal contiguity. Journal o f Experimental Analysis o f Behavior, 46, 15-35. Welch, B.L. (1951). On the comparison of several mean values: An alternative approach. Biometrika, 38, 330-336. Wood, R. E. (1986). Task complexity: Definition of the construct. Organizational Behavior and Human Decision Processes, 37, 60-82. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 Wood, R. E., Mento, A. J., & Locke, E. A. (1987). Task complexity as a moderator of goal effects: A meta-analysis. Journal o f Applied Psychology, 72, 416-25. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 Appendix A: Experimental Stimuli Condition 1: TAF, Reward SFM Alone Screen 1: Login (Same for All Conditions) Welcome to Lay Khim Ong’s Study Thank you for participating in this study. This study examines how managers make investment decisions. For the purpose of this study, you will assume the role of a division manager of a company. You will be paid a basic $5 for completing the study. In addition, you will be awarded points based on your performance in the study. These points will be converted to a cash reward at the end of the study. You will then be asked to provide your name and forwarding address and a check with the total cash you have earned will be mailed to you within 6 weeks. Please participate in this study only once and alone without the assistance of anyone. Your participation should take approximately an hour. Please ensure that you have a full block of about an hour to finish the entire study in one sitting. If you think you cannot finish the study in one sitting now, you may wish to exit the study and log in later. You can only log in once for this study. To begin, please log in with your user name and password. If you have not been assigned a user name and password, please contact Ms.Lay Khim Ong (email: ong@marshall.usc.edu). User name: Password: Submit Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Screen 2: User Validation (Same fo r All Conditions) Log in correct. Please continue. Continue Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 Screen 3: Introduction o f Ohm Inc. (Same for All Conditions) Introduction of Ohm Inc. You are a division manager in Ohm Inc., which is based in Santa Clara, California. Ohm Inc. designs, develops, manufactures, and markets a broad line of digital and mixed signal integrated circuits for a range of markets, including data communications, telecommunications, computers, and instrumentation systems. Since its founding, Ohm Inc. has developed 146 base products with over 750 variations. These products include interface products such as radio frequency devices that handle analog radio signals for digital data transmission; memory products such as static random access memory chips used for storage and retrieval of data in computers and other electronic systems, and programmable read-only memories; and timing technology products such as programmable clocks and PC clocks used for various functions such as telling time of day, recording elapsed time, and setting alarm timing. Ohm Inc. has four divisions and each division has its own wafer fabrication plant for manufacturing integrated circuits. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Screen 4: Task Description (Same for All Conditions) Your Task Ohm Inc. has a policy of rotating its division managers through all its four divisions. As a division manager, you will manage one division for a number of periods before being transferred to another division. You will not be informed in advance about the total number of periods that you will be managing each division. The four divisions are similar to and independent from each other. As such, you can use your experience in one division to help you manage the other divisions. In each period, you have a budget of $100m to invest in Research & Development (R&D) and/or Plant & Equipment (P&E). You have to spend the entire amount of $100m each period. R&D investments are used for developing new products to broaden the company's customer base, product lines, and end-product applications. P&E investments are used for building new wafer fabrication plants or installing new machinery and equipment to expand or upgrade manufacturing capabilities and improve operating efficiency. As division manager, your task is to allocate the $100m over R&D and/or P&E to maximize the operating income before deduction of R&D expenses and depreciation of each division. Operating income before deduction of R&D expenses and depreciation will hereafter be referred to as "gross operating income". Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Please answer the following questions with respect to the four divisions that you will be managing. For the purpose of these questions, gross returns refer to the additional gross operating income that a division earns as a result of making the R&D and/or P&E investment, without deducting the amount of the initial investment. For example, if R&D and/or P&E investment is $100m and the additional gross operating income generated is $150m, the gross returns would be $150m. What do you think are the gross returns of current period R&D investments in the current period? ONo returns O Small returns OMedium returns OLarge returns O Very large returns What do you think are the gross returns of current period R&D investments in future periods? ONo returns O Small returns OMedium returns OLarge returns OVery large returns What do you think are the gross returns of current period P&E investments in the current period? ONo returns O Small returns OMedium returns OLarge returns OVery large returns What do you think are the gross returns of current period P&E investments in future periods? ONo returns O Small returns OMedium returns OLarge returns OVery large returns Submit Reset Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Screen 6: Demographic Questionnaire (Same for All Conditions) Please answer the following questions about yourself: 138 Age Gender OMale O Female Number of years of full time work experience Your full time experience was most closely related to: O Accounting, Auditing, or Taxation OFinance, Banking, or Investing OS ales or Marketing O General Management O Human Resources O Engineering and Operations OResearch and Development O Other If you choose the option “Other”, please specify Number of financial accounting courses taken for your MBA degree to date: Number of management accounting courses taken for your MBA degree to date: Submit Reset Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Screen 7: Description o f Feedback Information Available (Same for Conditions 1 and 5, Differences from Other Conditions Shaded) Information Available to You At any period of a division, you will be able to view R&D investment, P&H investment, gross operating income, return on investment (ROl), and your points reward for that period and all prior periods. You can also view the same information for any division that you have already managed by clicking the appropriate division feedback button. ROI is calculated as follows: ROI = (Gross operating income - R&D expenses - Depreciation) * 100% Invested Amount Invested Amount = $100m R&D investment in each period (R&Dt) is immediately expensed in that period. R&D expenses in period t = R&Dt P&E investment in each period (P&Et) is depreciated using the straight-line me thod over an estimated useful life of 5 periods. Depreciation in period t = (P&Et)/5 + (P&Et_i)/5 + (P&Et_2)/5 + (P&Et.3)/5 + (P&Et. 4 )/5 Ohm Inc.’s choice of whether to expense or capitalize R&D and P&E investments is in accordance with generally accepted accounting principles and may not reflect actual economic reality. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 Screen 8: Description o f Compensation Scheme (Same fo r Conditions 1 and 5, Differences from Other Conditions Shaded) Your Compensation Scheme Each division has planned an ROI of 39% for each period. Each period, you will be compensated based on your actual versus planned performance on ROI. 2000 points are awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31 %). 2000 points are deducted for every % of achievement below 80% of planned ROI. Zero point is awarded if ROI is 80% of planned ROI. Points % of achievement above 80% of planned ROI 2000 1% 1% -2000 % of achievement above 80% of planned ROI Historical performance in each division indicates that 80% of planned ROI is an achievable target. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Screen 9: Prior Periods Investments (Same fo r All Conditions) Prior Periods Investments In the four periods before you begin managing each division, your predecessors have invested the following amounts in R&D and P&E: Period R&D Investments ($m) P&E Investments ($m) t-4 50 50 t-3 50 50 t-2 50 50 t-1 50 50 Each division is still learning about the impact of R&D and P&E investments on gross operating income and you are free to experiment with different levels of R&D and P&E investments to determine the optimal combination of R&D and P&E investments. As you manage each division, there will be no seasonal variations, external shocks, or unusual events that will change or hide the effects of R&D and P&E investments on gross operating income. Before you actually begin managing the four divisions, you will practice managing a division for 5 periods. The practice division will familiarize you with your task and the available feedback information. Your performance in the practice division will not be compensated. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 1 and 5 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income) Practice Division Prior Periods Results Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Points reward for ROI performance t-4 50 50 NA* NA* NA* t-3 50 50 NA* NA* NA* t-2 50 50 NA* NA* NA* t-1 50 50 NA* NA* NA* NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction of R&D expenses and depreciation. Practice Division, Period 1 You have a budget of $100m this year. How much do you want to invest in R&D and P&E? R&D ($m): P&E ($m): Enter Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 Screen 11: Feedback for Practice Division (Same for Conditions 1 and 5) Feedback Your Results for Practice Division, Period 1 Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Points reward for ROI performance t-4 50 50 NA* NA* NA* t-3 50 50 NA* NA* NA* t-2 50 50 NA* NA* NA* t-1 50 50 NA* NA* NA* 1 100 0 133.68 -6.32 -192410 NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction o f R&D expenses and depreciation. Next period (Screens 10 and 11 for Practice Division, Period 1 are repeated for Periods 2 to 5. Once the subject finishes managing the Practice Division, the subject is led to Screen 12 with the following message at the end of Screen 11.) You have finished the practice session. Please click 'Next' to begin managing your first division. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 Screen 12: Investments in R&D and P&E fo r Divisions (Same for Conditions 1 and 5 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income) Investments in R&D and P&E Prior Periods Results Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Points reward for ROI performance t-4 50 50 NA* NA* NA* t-3 50 50 NA* NA* NA* t-2 50 50 NA* NA* NA* t-1 50 50 NA* NA* NA* NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction of R&D expenses and depreciation. Division 1, Period 1 You have a budget of $100m this year. How much do you want to invest in R&D and P&E? R&D ($m): P&E ($m): Enter Your Compensation Scheme Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Screen 13: Feedback fo r Divisions (Same for Conditions 1 and 5) Feedback Your Results for Division 1, Period 1 Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Points reward for ROI performance t-4 50 50 NA* NA* NA* t-3 50 50 NA* NA* NA* t-2 50 50 NA* NA* NA* t-1 50 50 NA* NA* NA* 1 100 0 133.68 - 6.32 -192410 NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction of R&D expenses and depreciation. Next period (Screens 12 and 13 for Division 1, Period 1 are repeated for Periods 2 to 17. The screens are then repeated for Divisions 2 to 4. Division 1 has 17 periods. Division 2 has 15 periods. Division 3 has 15 periods. Division 4 has 14 periods. Once the subject finishes the last period of a division, the following message is displayed at the end of Screen 13.) You have finished managing Division 1. Continue Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 Screen 14: Completion o f Management o f Divisions (Same fo r All Conditions) Completion of Management of Division 1 You have earned a total of XXXXX points for Division 1. You will now be transferred to Division 2. Division 2 is similar to, but independent from the other three divisions of Ohm Inc. Since all four divisions are similar, information on any one division is relevant for helping you manage any of the four divisions. You may access Division 1 feedback by clicking the Division 1 feedback button. Next Division 1 feedback (Screen 14 is repeated for Division 2 and 3. Once the subject finishes managing Division 4, the subject is led to Screen 15 with the following message.) You have earned a total of XXXXX points for Division 4. You may access Division 4 feedback by clicking the Division 4 feedback button. Next Division 4 feedback Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 47 Screen 15: Post-Experimental Questionnaire, First Version (Same for All Conditions except Q3, Q3 Here Is the Same for Conditions 1 and 5) Your task as a division manager has now ended. Please answer all the following questions regarding the four divisions you have just managed: Ql) In what type of investments should you invest more to maximize the performance of a division across all periods? The performance of a division refers to the “gross operating income”, i.e., operating income before deduction of R&D expenses and depreciation. OR&D OP&E Q2) What are the relative weights (out of 100%) explicitly placed on achieving results in ROI, R&D productivity, and P&E operating efficiency in your compensation scheme? Enter “0” for any item that is not explicitly mentioned in your compensation scheme even if the item may indirectly affect your compensation. ROI ___ % R&D productivity % P&E operating efficiency ___ % Total 100 % Q3) You were given information on your performance on gross operating income and ROI. What was the relative attention (out of 100%) that you paid to these performance measures? Enter “0” for any performance measure that you did not pay attention to. Gross operating income % ROI " ~ 1% Total TOO % Q4) What was the relative attention (out of 100%) that you paid to the effects of R&D versus P&E investments on your performance? Enter ‘0’ for any item that you did not pay attention to. Effects of R&D investment on performance ___% Effects of P&E investment on performance ___% Total % Q5) Try to remember how you thought R&D investment affected your performance. Allocate 100 points over the following factors based on how important you thought they were when you assessed how R&D investment affected your performance. Enter “0” for any factor that you had not thought of. The factors listed below are not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 prescriptive or exhaustive. If you thought of other factors, please describe them in 5.4 5.1) Current R&D investment decreases current ROI because current R&D investment is immediately expensed. 5.2) Current R&D investment increases current gross operating income and current ROI because current R&D productivity increases as new products are introduced. 5.3) Current R&D investment increases future gross operating income and future ROI because current R&D productivity increases as new products are introduced. However, increased current R&D productivity takes some time before it positively affects gross operating income and ROI. 5.4) Other (Please describe below) Total 100 Please describe the other factors your thought of when you assessed how R&D investment affected your performance: Q6) Try to remember how you thought P&E investment affected your performance. Allocate 100 points over the following factors based on how important you thought they were when you assessed how P&E investment affected your performance. Enter “0” for any factor that you had not thought of. The factors listed below are not prescriptive or exhaustive. If you thought of other factors, please describe them in 6.5 6.1) Current P&E investment decreases current ROI because current P&E investment is capitalized and depreciated over 5 periods. 6.2) Current P&E investment decreases future ROI because current P&E investment is capitalized and depreciated over 5 periods. 6.3) Current P&E investment increases current gross operating income and current ROI because current P&E operating efficiency increases with new and better plant and machinery. 6.4) Current P&E investment increases future gross operating income and future ROI because current P&E operating efficiency increases with new and better plant and machinery. However, increased current operating efficiency takes some time before it positively affects gross operating income and ROI. 6.5) Other (Please describe below) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 Total 100 Please describe the other factors your thought of when you assessed how P&E investment affected your performance: Q7) Which type of investments generates larger returns to gross operating income (i.e. operating income before deduction of R&D expenses and depreciation) across all periods of a division? OR&D investment OP&E investment OR&D and P&E investments generate equal returns Q8) Consider the effects of increasing R&D investment on current period gross operating income. Select the statement that you agree the most with. OIncreasing R&D investment increases current period gross operating income. OIncreasing R&D investment decreases current period gross operating income. OIncreasing R&D investment does not affect current period gross operating income. Q9) Consider the effects of increasing R&D investment on future periods gross operating income. Select the statement that you agree the most with. OIncreasing R&D investment increases future periods gross operating income. OIncreasing R&D investment decreases future periods gross operating income. OIncreasing R&D investment does not affect future periods gross operating income. Q10) Consider the marginal returns on gross operating income of R&D investment, i.e., the change in returns per unit change in R&D investment. Select the statement that you agree the most with. OR&D investment generates diminishing marginal returns. For example, increasing R&D investment from $0m to $10m produces higher increase in returns than increasing R&D investment from $10m to $20m. OR&D investment generates constant marginal returns. For example, increasing R&D investment from $0m to $10m produces the same increase in returns as increasing R&D investment from $10m to $20m. OR&D investment generates increasing marginal returns. For example, increasing R&D investment from $0m to $10m produces lower increase in returns than increasing R&D investment from $10m to $20m. Q ll) Which is the best combination of R&D and P&E investments to maximize gross operating income across all periods of a division? OR&D: $20m P&E: $80m OR&D: $40m P&E: $60m OR&D: $50m P&E: $50m OR&D: $60m P&E: $40m Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 OR&D: $80m P&E: $20m Q12) Consider the effects of increasing P&E investment on current period gross operating income. Select the statement that you agree the most with. OIncreasing P&E investment does not affect current period gross operating income. OIncreasing P&E investment decreases current period gross operating income. OIncreasing P&E investment increases current period gross operating income. Q13) Consider the effects of increasing P&E investment on future periods gross operating income. Select the statement that you agree the most with. OIncreasing P&E investment does not affect future periods gross operating income. OIncreasing P&E investment decreases future periods gross operating income. OIncreasing P&E investment increases future periods gross operating income. Q14) Consider the marginal returns on gross operating income of P&E investment, i.e., the change in returns per unit change in P&E investment. Select the statement that you agree the most with. OP&E investment generates increasing marginal returns. For example, increasing P&E investment from $0m to $10m produces lower increase in returns than increasing P&E investment from $10m to $20m. OP&E investment generates constant marginal returns. For example, increasing P&E investment from $0m to $10m produces the same increase in returns as increasing P&E investment from $10m to $20m. OP&E investment generates diminishing marginal returns. For example, increasing P&E investment from $0m to $10m produces higher increase in returns than increasing P&E investment from $10m to $20m. Q15) Allocate 100 points over the following statements based on how well each statement describes the strategies that you used to understand the relationships between R&D/P&E investments and your performance. Enter “0” for any strategy that you did not use. The strategies listed below are not prescriptive or exhaustive. If you used other strategies, please describe them in 15.5 15.1) I vary the R&D and P&E investments in the current period from the prior period and assess the change in performance in the current period from the prior period. 15.2) I vary the R&D and P&E investments across periods. Then, I search for the combination of R&D and P&E investments in the period that produces the best performance in that same period. 15.3) I increase or decrease R&D and P&E investments consistently and gradually across periods and see the general trend of the performance. 15.4) I vary the R&D and P&E investments across periods. Then, I assess Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 the impact of current period R&D and P&E investments on future periods performance by seeing whether high/low R&D or P&E investments in the current period lead to high/low performance in future periods. 15.5) Other (Please describe below) Total 100 Please explain the other strategies that you used to understand the relationships between R&D/P&E investments and your performance. Submit Reset Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 Screen 16: Cash Reward (Same fo r All Conditions) Your Cash Reward For Division 1, you earned XXX points, which is converted to $Y.YY. For Division 2, you earned XXX points, which is converted to $Y.YY. For Division 3, you earned XXX points, which is converted to $Y.YY. For Division 4, you earned XXX points, which is converted to $Y.YY. For all four divisions, you have earned a total of SYY.YY. You have also earned $5 for completing the study. Please provide the following information so that the check with your cash reward can be mailed to you: Full name as it will appear on the check: Full mailing address: Submit Reset Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 Screen 17: End Message (Same for All Conditions) The End Thank you for your participation. A check with your cash reward will be forwarded to the address that you provided within 6 weeks. If you wish to provide any feedback regarding this study, you may email me at ong@marshall.usc. edu Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 Condition 2: TAF, Reward SFM Alone and Provide VD Information Screen 1: Log In (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same fo r All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Screen 7: Description o f Feedback Information Available (Same for Conditions 2, 3, 4, 6, 7 , and 8, Differences from Other Conditions Shaded) Information Available to You At any period of a division, you will be able to view the R&D investment, P&E investment, gross operating income, return on investment (ROI), R&D productivity, P&E operating efficiency, and your points reward for that period and all prior periods. You can also view the same information for any division that you have already managed by clicking the appropriate division feedback button. Each division measures R&D productivity by the number of new products, including new product applications, introduced in that period. Each division engages an external semiconductor consultant to audit its existing wafer fabrication facilities each period and rate the operating efficiency relative to industry standard on a scale of 1-100 (with 1 being the lowest and 100 being the highest). The R&D productivity and P&E operating efficiency are important information that will be useful for understanding the relationships between R&D/P&E investments and gross operating income and improving your performance. ROI is calculated as follows: ROI = (Gross operating income - R&D expenses - Depreciation) * 100% Invested Amount Invested Amount = $100m R&D investment in each period (R&Dt) is immediately expensed in that period. R&D expenses in period t = R&Dt P&E investment in each period (P&Et) is depreciated using the straight-line method over an estimated useful life of 5 periods. Depreciation in period t = (P&Et)/5 + (P&Et.i)/5 + (P&Et_2)/5 + (P&Et.3 )/5 + (P&Et. 4 )/5 Ohm Inc.’s choice of whether to expense or capitalize R&D and P&E investments is in accordance with generally accepted accounting principles and may not reflect actual economic reality. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 Screen 8: Description o f Compensation Scheme (Same fo r Conditions 2 and 6, Differences from Other Conditions Shaded) Your Compensation Scheme Each division has planned an ROI of 39%, 46 new products, and an operating efficiency of 92 for each period. Each period, you will be compensated based on your actual versus planned performance on ROI. 2000 points are awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31%). 2000 points are deducted for every % of achievement below 80% of planned ROI. Zero point is awarded if ROI is 80% of planned ROI. Points % o f achievement above 80% of planned ROI 2000 1% 1% -2000 % of achievement above 80% of planned ROI Historical performance in each division indicates that 80% of planned ROI is an achievable target. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 Screen 9: Prior Periods Investments (Same for All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 2, 3, 6 and 7 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 2, 3, 6 and 7 , Similar to Screen 13) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 Screen 12: Investments in R&D and P&E for Divisions (Same fo r Conditions 2, 3, 6 and 7 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income) Investments in R&D and P&E Prior Periods Results Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Number o f new products Operating efficiency rating Points reward for ROI performance t-4 50 50 NA* NA* NA* NA* NA* t-3 50 50 NA* NA* NA* NA* NA* t-2 50 50 NA* NA* NA* NA* NA* t-1 50 50 NA* NA* NA* NA* NA* NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction of R&D expenses and depreciation. Division 1, Period 1 You have a budget of $100m this year. How much do you want to invest in R&D and P&E? R&D ($m): P&E ($m): Enter Your Compensation Scheme Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 Screen 13: Feedback for Divisions (Same for Conditions 2, 3, 6 and 7) Feedback Your Results for Division 1, Period 1 Period R&D Investment ($m) P&E Investment ($m) Gross operating income ($m) ROI (%) Number of new products Operating efficiency rating Points reward for ROI performance t-4 50 50 NA* NA* NA* NA* NA* t-3 50 50 NA* NA* NA* NA* NA* t-2 50 50 NA* NA* NA* NA* NA* t-1 50 50 NA* NA* NA* NA* NA* 1 100 0 133.68 -6.32 46 0 -192410 NA* Information not available for Periods t-4, t-3, t-2, and t-1. Gross operating income is operating income before deduction of R&D expenses and depreciation. Next period (Screens 12 and 13 for Division 1, Period 1 are repeated for Periods 2 to 17. Once the subject finishes managing Division 1, the screens are then repeated for Divisions 2-4.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 Screen 14: Completion o f Management o f Divisions (Same for All Conditions) Screen 15: Post-Experimental Questionnaire (Same fo r All Conditions except Q3, Q3 Here Is the Same for Conditions 2, 3, 4, 6, 7 , and 8) Q3) You were given information on your performance on gross operating income, ROI, R&D productivity (number of new products), and P&E operating efficiency (operating efficiency rating). What was the relative attention (out of 100%) that you paid to these performance measures? Enter “0” for any performance measure that you did not pay attention to. Gross operating income _ % ROI R&D productivity P&E operating efficiency Total 100% Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 Condition 3: TAF, Reward SFM Alone and Provide VD information and VD weights Screen 1: Login (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same fo r All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Screen 7: Description o f Feedback Information Available (Same for Conditions 2, 3, 4, 6, 7 , and 8, See Condition 2.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 Screen 8: Description o f Compensation Scheme (Differences from Other Conditions Shaded) Your Compensation Scheme Ohm inc. believes that the performance of each division is based 43.5% on ROI, 13% on R&D productivity, and 43.5% on P&E operating efficiency rating. Each division has planned an ROI of 39%, 46 new products, and an operating efficiency of 92 for each period. Each period, you will be compensated based on your actual versus planned performance on ROI. 2000 points are awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31%). 2000 points are deducted for every % of achievement below 80% of planned ROI. Zero point is awarded if ROI is 80% of planned ROI. P o in ts % of achievement above 80% of planned ROI 2000 1% 1% -2000 % of achievement above 80% of planned ROI Historical performance in each division indicates that 80% of planned ROI is an achievable target. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 Screen 9: Prior Periods Investments (Same fo r All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 2, 3, 6 and 7 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 2, 3, 6 and 7 , Similar to Screen 13) Screen 12: Investments in R&D and P&E (Same for Conditions 2, 3, 6 and 7 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, See Condition 2) Screen 13: Feedback for Divisions (Same fo r Conditions 2, 3, 6 and 1 , See Condition 2 ) Screen 14: Completion o f Management o f Divisions (Same fo r All Conditions) Screen 15: Post-Experimental Questionnaire (See Condition 2) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 Condition 4: TAF, Reward SFM and Accurately-Weighted VDs Screen 1: Login (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same fo r All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior Beliefs Questionnaire (Same fo r All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Screen 7; Description o f Feedback Information Available (Same for Conditions 2, 3, 4, 6, 7 , and 8, See Condition 2) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 Screen 8: Description o f Compensation Scheme (Differences from Other Conditions Shaded) Your Compensation Scheme Ohm Inc. believes that the performance of each division is based 43.5% on ROI, 13% on R&D productivity, and 43.5% on P&E operating efficiency rating. Each division has planned an ROI of 39%, 46 new products, and an operating efficiency of 92 for each period. Each period, you will be compensated based on your actual versus planned performance on ROI, R&D productivity (number of new products), and P&E operating efficiency (operating efficiency rating). You will be rewarded as follows: ROI Number of new products Operating efficiency rating 1850 points awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31%). 1850 points deducted for every % of achievement below 80% of planned ROI. Zero point awarded for 80% of planned ROI. 550 points awarded for every % of achievement above 80% of planned number of new products (i.e., 37 new products). 550 points deducted for every1 % of achievement below 80% of planned number of new products. Zero point awarded for 80% of planned number of new products. 1850 points awarded for every % of achievement above 80% of planned operating efficiency rating (i.e., an operating efficiency rating of 74). 1850 points deducted for every % of achievement below 80% of planned operating efficiency rating. Zero point awarded for 80% of planned operating efficiency rating. Points % of achievement above 80% of planned ROI 1850 1% 1% - 1850 % of achievement above 80% of planned ROI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 Points % of achievement above 80% of planned number of new products 550 1% 1% -550 % of achievement above 80% of planned number of new products Points % of achievement above 80% of planned operating efficiency rating 1850 1% 1% -1850 % of achievement above 80% of planned operating efficiency rating Historical performance in each division indicates that 80% of planned ROI, number of new products, and operating efficiency rating are achievable targets. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 Screen 9: Prior Periods Investments (Same fo r All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 4 and 8 Except for the Underlying Algorithm Linking R&D and P&E investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback fo r Practice Division (Same for Conditions 4 and 8, Similar to Screen 13) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 Screen 12: Investments in R&D and P&E (Same for Conditions 4 and 8 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income) Investments in R&D and P&E Prior Periods Results Period R& D Invest -m ent ($m ) P& E Invest -m ent ($m ) Gross operating incom e ($m ) ROI (%) N um ber o f n ew products O perating efficien cy rating Points reward for ROI perform ance P oints reward for R & D productivity perform ance P oin ts reward for P& E operating efficien cy perform ance T otal points reward t-4 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-3 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-2 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-1 50 50 N A * N A * N A * N A * N A * N A * N A * N A * N A * Inform ation not available for Periods t-4, t-3, t-2, and t-1. G ross operating incom e is operating incom e before deduction o f R & D exp en ses and depreciation. Division 1, Period 1 You have a budget of $100m this year. How much do you want to invest in R&D and P&E? R&D ($m): P&E ($m): Enter Your Compensation Scheme Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 Screen 13: Feedback fo r Divisions (Same fo r Conditions 4 and 8) Feedback Your Results for Division 1, Period 1 Period R& D Invest -ment ($m ) P& E Invest -m ent ($m ) G ross operating incom e ($m ) ROI (%) N um ber o f n ew products Operating efficien cy rating P oints reward for ROI perform ance Points reward for R & D productivity perform ance P oints reward for P& E operating efficien cy perform ance Total points reward t-4 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-3 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-2 50 50 N A * N A * N A * N A * N A * N A * N A * N A * t-1 50 50 N A * N A * N A * N A * N A * N A * N A * N A * 1 100 0 133.68 - 6.32 46 0 -177979 11000 -1 4 8 0 0 0 -314979 N A * Inform ation not available for P eriods t-4, t-3, t-2, and t-1. Gross operating incom e is operating incom e before deduction o f R & D exp en ses and depreciation. Next period (Screens 12 and 13 for Division 1, Period 1 are repeated for Periods 2 to 17. Once the subject finishes managing Division 1, the screens are then repeated for Divisions 2-4.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 Screen 14: Completion o f Management o f Division 1 (Same fo r All Conditions) Screen 15: Post-Experimental Questionnaire (See Condition 2) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 Condition 5: ITAF, Reward SFM Alone Screen 1: Login (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same for All Conditions) Screen 4: Task Description (Same fo r All Conditions) Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Screen 1: Description o f Feedback Information Available (Same fo r Conditions 1 and 5, See Condition 1) Screen 8: Description o f Compensation Scheme (Same for Conditions 1 and 5, See Condition 1) Screen 9: Prior Periods Investments (Same for All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 1 and 5 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 1 and 5, Similar to Screen 13) Screen 12: Investments in R&D and P&E (Same for Conditions 1 and 5 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, See Condition 1) Screen 13: Feedback for Divisions (Same fo r Conditions 1 and 5, See Condition 1) Screen 14: Completion o f Management o f Divisions (Same for All Conditions) Screen 15: Post-Experimental Questionnaire (Same for All Conditions Except Q3, See Condition 1) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 Condition 6: ITAF, Reward SFM Alone and Provide VD information Screen 1: Log In (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same for All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Screen 7; Description o f Feedback Information Available (Same for Conditions 2, 3, 4, 6, 7 , and 8, See Condition 2) Screen 8: Description o f Compensation Scheme (Same for Conditions 2 and 6, See Condition 2) Screen 9: Prior Periods Investments (Same for All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 2, 3, 6 and 7 Except fo r the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 2, 3, 6 and 7 , Similar to Screen 13) Screen 12: Investments in R&D and P&E (Same for Conditions 2, 3, 6 and 7 Except for the Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, See Condition 2) Screen 13: Feedback fo r Divisions (Same for Conditions 2, 3, 6 and 7) Screen 14: Completion o f Management o f Divisions (Same for All Conditions) Screen 15: Post Experimental Questionnaire (Same for All Conditions Except Q3, See Condition 2) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 Condition 7: ITAF, Reward SFM and Provide VD information and VD weights Screen 1: Login (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same for All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same for All Conditions) Screen 7: Description o f Feedback Information Available (Same for Conditions 2, 3, 4, 6, 7 , and 8, See Condition 2.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 Screen 8: Description o f Compensation Scheme (Differences from Other Conditions Shaded) Your Compensation Scheme Ohm Inc. believes that the performance of each division is based 43.5% on ROI, 43.5% on R&D productivity, and 13% on P&E operating efficiency rating. Each division has planned an ROI of 39%, 46 new products, and an operating efficiency of 92 for each period. Each period, you will be compensated based on your actual versus planned performance on ROI. 2000 points are awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31%). 2000 points are deducted for every % of achievement below 80% of planned ROI. Zero point is awarded if ROI is 80% of planned ROI. Points % of achievement above 80% of planned ROI 2000 1% 1% -2000 % of achievement above 80% of planned ROI Historical performance in each division indicates that 80% of planned ROI is an achievable target. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 Screen 9: Prior Periods Investments (Same for All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 2, 3, 6 and 7 Except for Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 2, 3, 6 and 7 , Similar to Screen 13.) Screen 12: Investments in R&D and P&E (Same for Conditions 2, 3, 6 and 7 Except for Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, See Condition 2) Screen 13: Feedback for Divisions (Same for Conditions 2, 3, 6 and 7 , See Condition 2.) Screen 14: Completion o f Management o f Divisions (Same fo r All Conditions) Screen 15: Post Experimental Questionnaire (Same for All Conditions Except Q3, See Condition 2) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Condition 8: ITAF, Reward SFM and Accurately-Weighted VDs Screen 1: Login (Same for All Conditions) Screen 2: User Validation (Same for All Conditions) Screen 3: Introduction o f Ohm Inc. (Same for All Conditions) Screen 4: Task Description (Same for All Conditions) Screen 5: Prior-Beliefs Questionnaire (Same for All Conditions) Screen 6: Demographic Questionnaire (Same fo r All Conditions) Screen 7: Description o f Feedback Information Available (Same fo r Conditions 2, 3, 4, 6, 7 , 8, See Condition 2.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Screen 8: Description o f Compensation Scheme (Differences from Other Conditions Shaded) Your Compensation Scheme Ohm Inc. believes that the performance of each division is based 43.5% on ROI, 43.5% on R&D productivity, and .13% on P&E operating efficiency rating. Each division has planned an ROI of 39%, 46 new products, and an operating efficiency of 92 for each period. Each period, you will be compensated based on your actual versus planned performance on ROI, R&D productivity (number of new products), and P&E operating efficiency (operating efficiency rating). You will be rewarded as follows: ROI Number of new products Operating efficiency rating 1850 points awarded for every % of achievement above 80% of planned ROI (i.e., an ROI of 31%). 1850 points deducted for every % of achievement below 80% of planned ROI. Zero point awarded for 80% of planned ROI. 1850 points awarded for every % of achievement above 80% of planned number of new products (i.e., 37 new products). 550 points deducted for every % of achievement below 80% of planned number of new products. Zero point awarded for 80% of planned number of new products. 550 points awarded for every % of achievement above 80% of planned operating efficiency rating (i.e., a rating of 74). 1850 points deducted for every % of achievement below 80% of planned rating. Zero point awarded for 80% of planned operating efficiency rating. Points % of achievement above 80% of planned ROI 1850 1% 1% -1850 % of achievement above 80% of planned ROI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 Points % of achievement above 80% of planned number of new products 1% 1% - 1850 % of achievement above 80% of planned number of new products Points % of achievement above 80% of planned operating efficiency rating 1% 1% -550 % of achievement above 80% of planned operating efficiency rating Historical performance in each division indicates that 80% of planned ROI, number of new products, and operating efficiency rating are achievable targets. The points awarded will be accumulated across all periods for each division. At the end of the study, positive total points for a division will be proportionately converted to a cash reward while negative total points for a division will be converted to zero cash reward. Next Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 Screen 9: Prior Periods Investments (Same for All Conditions) Screen 10: Investments in R&D and P&E for Practice Division (Same for Conditions 4 and 8 Except fo r Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, Similar to Screen 12) Screen 11: Feedback for Practice Division (Same for Conditions 4 and 8, Similar to Screen 13) Screen 12: Investments in R&D and P&E (Same for conditions 4 and 8 Except for Underlying Algorithm Linking R&D and P&E Investments to Gross Operating Income, See Condition 4) Screen 13: Feedback for Divisions (Same for Conditions 4 and 8, See Condition 4) Screen 14: Completion o f Management o f Divisions (Same fo r All Conditions) Screen 15: Post Experimental Questionnaire (See Condition 2) Screen 16: Cash Reward (Same for All Conditions) Screen 17: End Message (Same for All Conditions) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 Appendix B: Properties of the Natural Log Function A natural log function is used to relate R&Dt to R&DPt and P&Et to P&EOEt because it has several useful properties. R&DPt = 10 ln(R&Dt+ 1). P&EOEt = 20 ln(P&Et +1). 1. R&Dt and P&Et have positive marginal rates of return to R&DPt and P&EOEt, respectively. cfR&DPt/cfR&Dt > 0 dP&EOEt/dP &E t > 0 2. R&Dt and P&Et have diminishing marginal rates of return to R&DPt and P&EOEt, respectively, reflecting reality. /R & D P t/r/R&Dt2 < 0 /P&EOEt/rfP&Et2 < 0 3. ln(R&Dt + 1) is used rather than ln(R&Dt) and ln(P&Et + 1) is used rather than ln(P&Et). This is because l n x < 0 i f 0 < x < l;lnx = 0ifx = l; and In x is undefined if x = 0. For example, if ln(R&Dt) is used, 0 < R&Dt < 1 will result in negative returns to R&DPt, R&Dt = 1 will result in zero returns to R&DPt, and R&Dt = 0 will produce undefined returns to R&DPt. Using ln(R&Dt + 1 ) rather than ln(R&Dt) and ln(P&Et + 1) rather than ln(P&Et) shifts the log function curves to the left by 1 such that all R&D and P&E investments impact the number of new product introductions and operating efficiency rating, respectively, positively; and that a zero Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 R&D and P&E investment has zero impact on the number of new product introductions and operating efficiency rating, respectively. A natural log function, a form of Cobb-Douglas function1 , is used to relate R&Dt - 3 and P&Et - 3 to 7tt because it has several useful properties (Varian, 1996). 7tt = a ln(R&Dt - 3 + 1) + b ln(P&Et . 3 + 1 ). 1. The isoquants are convex monotonic curves, that is, well-behaved indifference curves. In my study, an isoquant plots all combinations of R&Dt _ 3 and P&Et_ 3 that generate the same nt. The Cobb-Douglas function is the simplest algebraic expression that generates well-behaved preferences. Well-behaved indifference curves have various important and useful properties. First, preferences are monotonic, that is, more of either R&Dt - 3 or P&Et- 3 generates larger 7tt. The indifference curve, therefore, have a negative slope. Second, preferences are convex and strictly convex, that is, average combinations of R&Dt - 3 and P&Et _ 3 are strictly preferred to extreme combinations of R&Dt - 3 and P&Et_ 3 . 2. R&Dt - 3 and P&Et _ 3 have positive marginal rates of return to 7tt. dnt/dK8cDt.3 > 0 dnt/dP8cEt.2 > 0 3. R&Dt _ 3 and P&Et - 3 have diminishing marginal rates of return to 7 tt, reflecting reality. 1 A Cobb-Douglas function has the form f(xhx2) = Ax]a x2 b. It has been used for describing production functions as well as utility functions. In a production function, A represents the scale of production, that is, the amount of output if one unit of each input is used, a and b represent how the amount of output will change with changes in inputs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 82 J 27rt/ < f R & D t-32 < 0 d2%tld£&Ei.i2 < 0 4. R&Dt - 3 and P&Et _ 3 are not perfect substitutes for each other, reflecting reality. Managers have to balance their investments in R&D and P&E, taking into account the relative impact of each investment on in order to optimize 7tt. 5. R&Dt - 3 and P&Et _ 3 can be weighted differently to reflect their relative impact on nt, that is, they can have different coefficients. 6 . ln(R&Dt - 3 + 1) is used rather than ln(R&Dt.3 ) and ln(P&Et- 3 + 1 ) is used rather than ln(P&Et- 3 ). This is because In x < 0 if 0 < x < 1; In x = 0 if x = 1; and In x is undefined if x = 0. For example, if ln(R&Dt- 3 ) is used, 0 < R&Dt - 3 < 1 will result in negative returns to 7tt, R&Dt _ 3 = 1 will result in zero returns to nt, and R&Dt - 3 = 0 will produce undefined returns to 7 tt. Using ln(R&Dt - 3 + 1) rather than ln(R&Dt- 3 ) and ln(P&Et - 3 1 ) rather than ln(P&Et_3 ) shifts the log function curves to the left by 1 such that all R&D and P&E investments impact 7tt positively and that a zero R&D and P&E investment has zero impact on nt. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 Appendix C: Calculations of Optimal R&Dt- 3 , P&Et„ 3 , and Tangible Assets Firm (TAF) % t = 8 ln(R&Dt - 3 + 1) + 26 ln(P&Et - 3 + 1) T C t was maximized when dittldR&Dt-3 — dnt/dP&Et.3, otherwise 7x t could be increased by investing in the investment with the larger marginal return. fi?7ctM R & D t-3 = 8 / ( R & D t-3 + 1) dud dP &Et-3 - 26/(P&Et-3 + 1 ) 8 /(R&Dt _ 3 + 1) = 26/(P&Et.3 + 1) (R&Dt . 3 + l)/(P&E t . 3 + 1) = 8/26 (1) Since the budget for investments in each period was fixed at $100 million and the entire $ 1 0 0 million had to be spent each period, R&Dt . 3 + P&Et - 3 = 100 (2) Solving the two simultaneous equations: Optimal R&Dt - 3 = 23 Optimal P&Et - 3 = 77 Optimal 7it = 8 ln(23 + 1) + 26 ln(77 + 1) = $138.70 million Intangible Assets Firm (ITAF) 7tt - 26 ln(R&Dt - 3 + 1) + 8 ln(P&Et - 3 + 1) 7tt is maximized when d7tt/dR&Dt . 3 = 6 ? 7c t/<iP&Et.3 . dnJdR&DM = 26/(R&Dt-3 + 1) dK\! c/P&Et- 3 = 8 /(P&Et - 3 + 1) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 8 4 26/(R&Dt.3 + 1) = 8 /(P&Et - 3 + 1) (R&Dt - 3 + l)/(P&E t . 3 + 1) = 26/8 (3) R&Dt _ 3 + P&Et. 3 = 100 (4) Solving the two simultaneous equations: Optimal R&Dt - 3 = 77 Optimal P&Et- 3 = 23 Optimal 7it = 26 ln(77 + 1) + 8 ln(23 + 1) = $138.70 million Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 Appendix D: Calculations of Expected 7tt, R&Dt, P&Et, R&DPt, and P&EOEt E(7tt) for TAF T it ~ 8 ln(R&Dt - 3 + 1) + 26 ln(P&Et - 3 + 1) Let R&Dt - 3 = R and P&Et _ 3 = P Since R + P = 100, then rct = 8 ln(R + 1) + 26 ln(101 - R) Assume that the probability density function of R,/(R) = 1/100, a uniform density on [0 , 1 0 0 ], <400 <400 E(7T t) = | 8 ln(R + 1)/(R )dR + | 26ln(l01 - R )/(R )dR = (8/100)|°°ln(R +1) dR + (26/100)|°°ln(101 - R) dR ! = (8/100)[(R +1 )ln(R +1) - r]q 0 0 + (26 /100)[R ln(l 0 1 - R )-101 ln(l 01 - R) - R]q 0 0 - (8 / 1 0 0 )[( 1 0 1 1 n l 0 1 - 1 0 0 ) - lnl] + (26/100)[(1001nl - lOllnl - 100) - (- lOllnlOl)] = (8 /100)( 101 lnl 01 - 100) + (26/1 Q0)(- 100 + lOllnlOl) = 124.48 To solve lln(R + 1) dR using integration by parts where: If(x)g(x) dx = f(x)g(x) - 1 f(x)g’(x) dx Let f (R)g(R) = ln(R +1) f(R) = R + 1 f (R) = 1 g(R) = ln(R+ 1) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 g’(R) = 1/(R + 1) f ln(R + l)<fR = f(R)g(R) - J f(R)g’(R) dR = (R+ l)ln(R + 1) - j(R+l)[l/(R+l)] JR = (R + l)ln(R + 1) - R To solve j ln(101 - R) dR using integration by parts: Let f (R)g(R) = ln(101 - R) f(R) = -(101 — R) = R - 101 f (R) = 1 g(R) = ln(101 -R ) g’(R) = - 1/(101 - R ) = 1/(R - 101) \ ln(101 -R ) d R = f(R)g(R) - 1 f(R)g’(R) dR = (R - 101)ln(101 - R) - j (R - 101)[1/(R - 101)] dR = Rln(101 - R) - 1011n(101 - R ) - R E(m) for ITAF T it = 26 ln(R&Dt - 3 + 1) + 8 ln(P&Et . 3 + 1 ) = 26 ln(R + 1 ) + 8 ln(101 - R) Assume that the probability density function of R ,fiR ) = 1/100, a uniform density on [0 , 1 0 0 ], A 00 ,400 E(7it) = I 26ln(R + 1)/(R )dR+ J 8 ln(l01 - R )/(R )dR (too (too - (26/100)| ln(R + l)dR + (8 / 1 0 0 ) J ln(101-R)<fR Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 - (26/100)[(R + l)ln(R + 1 ) - R ] r + ( 8 /100)[R ln(l 01 - R) -101 ln(l 01 - R) - R]q°° = (26/100)[(1011nl01 - 1 0 0 ) - lnl] + (8/100)[(1001nl - lOllnl - 100) - (- lOllnlOl)] = (26/100)[1011nl01 - 100] + (8/100)[- 100 + lOllnlOl] = 124.48 E(R&Dts) Let R&Dt - 3 = R j 400 E ( R & D t-3) = | Rf(R)dR AOO = (1/100) | RdR = (l/ 1 0 0 )[RV 2 t " = 1 0 0 0 0 / 2 0 0 = 50 E(R&DPt-s) R&DPt-s = 10 ln(R&Dt . 3 + 1 ) = 10 ln(R + 1) *400 E(R&DPt-3 )= | 101n(R + l)/(R )rfR <400 = 10/100 | ln(R + l) c/R = 10/100 [(R +1) ln(R +1) - R f = 1 0 / 1 0 0 [(1 0 1 1 n l 0 1 - 1 0 0 ) - In 1 ] = 36.61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 188 = 37 (rounded to the nearest figure) E(P&Et-s) P&Et - 3 = 100 - R (400 E(P&Et.3 ) = | (100-R )/(R )t/R A 00 = (1/100) I (100-R) dR. = (1/100) [l 00R - (R 2 / 2) }0 ° ° = (1/100) [10000 - (10000/20] = 50 E(P&EOEt-s) P&EOEt - 3 - 20 ln(P&Et.3 +l) = 20 ln(101 - R) E(P&EOEt_ 3 ) = | 201n(l01 - R )/(R ) < 7 R A 00 = 20/100 | ln(101-R)i/R = 20/100 [R ln(l 01 - R) -101 ln(l 01 - R) - R]o °° = 20/100[(1001nl - lOllnl - 100) - (- lOllnlOl)] = 2 0 / 1 0 0 [ 1 0 1 1 n l 0 1 - 1 0 0 ] = 73.23 = 73 (rounded to the nearest figure) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189 Appendix E: Calculation of Expected Points Reward for Each Condition Condition 1, Condition 2, and Condition 3 Let the points awarded in period t of division k for performance be P*. Each period, 2 0 0 0 points were awarded (deducted) for every percent of achievement above (below) 80% of planned ROI*. Zero point was awarded if ROItk was 80% of planned ROItk- Planned ROItk = Optimal ROI* = 38.70% = 39% Optimal ROItk = (Optimal 7t* - Optimal R&D* - Optimal P&Etk - 4 / 5 - Optimal P&Etk-3 / 5 - Optimal P&Etk - 2 / 5 - Optimal P&Etk-i/5 - Optimal P&Etk/5)/100 = [138.70 - 23 - 77/5 - 77/5 - 77/5 - 77/5 - 77/5J/100 = 0.3870 = 38.70% E(Ptk ) = E{[(ROW39) - 0.80]*100*2000} = 5128.2051 E(ROItk) - 160000 For t = 1 , 2 , and 3 ,7i* = 8 ln(51) + 26 ln(101 - 50) = 133.68 since R&D* and P&E* were fixed at $50 million each for the four periods prior to the first period. E(ROIik ) - E[7tik - R&Dlk - (100 - R&D_4k)/5 - (100 - R&D_3 k)/5 - (100 - R&D_2k)/5 - (100 - R&D.ik)/5 - (100 - R&Dik )/5] = E(133.68 - R&Dik - 100 + 4(50/5) + R&Dlk/5) 400 = 73.68 + 1 - (4/5) R&Dik/[R&Dik ) t/R&Dlk = 73.68-4/5*(50) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 90 = 33.68 Therefore, E(P3 k ) = 5128.2051*33.68 - 160000 - 12718 E(ROI2 k ) = E[7i2 k - R&D2 k - (100 - R&D.3 k )/5 - ( 1 0 0 - R&D.2 k )/5 - (100 - R&D-ik )/5 - (100 - R&Dlk)/5 - (100 - R&D2 k )/5] = E(133.68 - R&D2 k - 100 + 3(50/5) + R&Dlk/5 + R&D2 k /5) ,400 (ioo _ = 63.68 + | - (4/ 5) R&D2 k /(R&D2 k ) c/R&D2 k + J (1/5)R&Di^R&Djk) dK&Dxk = 63.68 - 4/5(50) + 1/5(50) = 33.68 Therefore, E(P2 k ) = 5128.2051*33.68 - 160000 = 12718 E(ROI3 k ) = E[7t3 k - R&D3 k - (100 - R&D_2 k )/5 - (100 - R&D.lk)/5 - (100 - R&Dlk)/5 - (100 - R&D2 k )/5 - (100 - R&D 3 k )/5] = E(133.68 - R&D3 k - 100 +2(50/5) + R&Dik /5 + R&D2 k /5 + R&D 3 k /5) <400 (400 = 53.68 + | - (4/5) R&D3 k /(R&D3 k ) r/R&D3 k + | (l/5)R& D lk/(R&Dik ) dR&DX k + | ° ° ( l / 5 ) R & D 2k/ ( R & D 2 k) r/R&D2 k = 53.68 - 4/5(50) + 1/5(50) + 1/5(50) = 33.68 Therefore, E(P3 k ) = 5128.2051*33.68 - 160000 = 12718 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. For t > 4, E(7itk) = 124.48 (see Appendix D for calculations) 191 E(ROl4 k) = E[7i4 k - R&D4 k - (100 - R&D.ik)/5 - (100 - R&Dik)/5 - (100 - R&D2 k)/5 - ( 1 0 0 - R&D3 ic)/5 - (100 - R&D 4 k )/5] = E(124.48 - R&D4 k - 100 + (50/5) + R&Dik/5 + R&D2 k /5 + R&D3 k /5 + R&D4 k /5) (400 (400 = 34.48 + | - (4/ 5) R&D4 k /(R&D4 k ) dR&D4 k + J (l/5)R&Dm/(R&Dik ) dR&Dlk (400 ,400 + | (1/5) R&D2 k /(R&D2 k ) <fR&D2 k + J (1/5) R&D 3 k /(R&D3 k ) JR&D3 k = 34.48 - 4/5(50) + 1/5(50) + 1/5(50) + 1/5(50) = 24.48 Therefore, E(P4 k ) - 5128.2051*24.48 - 160000 = -34462 Since E(P5 k ) to E(Piastt,k) are identical, for E(P5 k ) to E(PiasU ,k ): E(ROI5 k ) = E[7i5 k - R&D5 k - (100 - R&Dik )/5 - (100 - R&D2 k )/5 - (100 - R&D3 k )/5 - (100 - R&D4 k )/5 - (100 - R&D5 k )/5] = E(124.48 - R&D5 k - 100 + R M V 5 + R&D2 k /5 + R&D3 k /5 + R&D4 k /5 + R&D5k /5 ,400 (400 = 24.48 + | ~(4/5)R & D 5 k /(R&D5 k ) c/R&D5 k + | (l/5)R& Dik /(R&Dlk) dR&Dik (400 <400 + | (1/5) R&D2 k /(R&D2 k ) c?R&D2 k + | (1/5) R&D3 k /(R&D3 k ) dR&D3 k ,400 + (1/5) R&D4 k /(R&D4 k ) <fR&D4 k = 24.48 - 4/5(50) + 1/5(50) + 1/5(50) + 1/5(50) + 1/5(50) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 = 24.48 Therefore E(P5 k ) = 5128.2051*24.48 - 160000 = - 34462 E(Total Points for Division 1 with 17 periods) = 12718*3 + (- 34462)* 14 = - 444314 E(Total Points for Division 2 with 15 periods) = 12718*3 + (- 34462)* 12 = -375390 E(Total Points for Division 3 with 15 periods) = 12718*3 + (- 34462)* 12 = -375390 E(Total Points for Division 4 with 14 periods) = 12718*3 + (- 34462)* 11 = - 340928 E(Total Points for all 4 divisions) = - 1536022 Condition 4 Let the points awarded in period t of division k for performance be Ptk, the points awarded in period t of division k for ROItk performance be PROItk, the points awarded in period t of division k for R&DPtk performance be PR&Dtk , and the points awarded in period t of division k for P&EOEtk performance be PP&Etk. Each period, 1850 points were awarded (deducted) for every percent of achievement above (below) 80% of planned ROItk. Zero point was awarded if ROItk was 80% of planned ROItk. 550 points were awarded (deducted) for every % of achievement above (below) 80% of planned number of new products, R&DPtk. Zero point was awarded if R&DPtk was 80% of planned number. 1850 points were awarded (deducted) for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 193 every % of achievement above (below) 80% of planned operating efficiency rating, P&EOEtk- Zero point was awarded if P&EOEtk was 80% of planned rating. Planned ROItk = Optimal ROItk = 38.70% = 39% E(PROItk ) = E{[(ROItk /39) - 0.80]*100*1850} = 4743.5 897*E(ROItk ) - 148000 For t=l, 2 , and 3 ,7 itk = 8 ln(51) + 26 In (101 - 50) = 133.68 since R&Dtk and P&Etk were fixed at $50m each for the four periods prior to the first period. E(ROIik) = 33.68 (see Conditions 1 , 2, and 3 for calculations) Therefore, E(PROIik ) = 4743.5897*33.68 - 148000 - 11764 Planned R&DPtk - Maximum R&DPtk = 10 ln(100 + 1) = 46 E(PR&Dtk ) = E{[(R&DPtk /46) - 0.80]* 100*550} = 1195.6522*E(R&DPtk) - 44000 E(R&DPik ) = 36.61 (see Appendix D for calculations) Therefore, E(PR&Dik ) = 1195.6522*36.61 - 44000 - - 224 Planned P&EOEtk = Maximum P&EOEtk = 20 ln(100 + 1) - 92 E(PP&Etk) = E{[(P&EOEtk /92) - 0.80]* 100*1850)} = 2010.8696*E(P&EOEtk ) - 148000 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 194 E(P&EOEik) = 73.23 (see Appendix D for calculations) Therefore, E(PP&Eik) = 2010.8696*73.23 - 148000 = - 753 Since E(Plk) = E(PROIik) + E(PR&Dlk) + E(PP&Eik), E(Pik) - 11764 - 224 - 753 = 10787 E(Plk) = E(P2 k ) = E(P3 k ) = 10787 For t > 4, E(7itk) - 124.48 (see Appendix D for calculations) E(ROl4 k) = 24.48 (see Conditions 1 , 2 , and 3 for calculations) E(PROLtk) = 4743.5897*24.48 - 148000 = - 31877 E(PR&D4 k ) = -2 2 4 E(PP&E4 k ) = - 753 Therefore, E(P4 k ) = - 31877 - 224 - 753 = - 32854 Since E(PROl5 k ) to E(PROIia stt, k) are identical, for E(PROl5 k ) to E(PROIia stt,k ): E(ROl5 k) = 24.48 (see Conditions 1, 2, and 3 for calculations) E(PROI5 k ) = 4743.5897*24.48 - 148000 = - 31877 E(PR&D5 k ) = -2 2 4 E(PP&E5 k ) = -753 Therefore, E(P5 k ) = -3 1 8 7 7 -2 2 4 - 753 = - 32854 E(Total Points for Division 1 with 17 periods) = 10787*3 + (- 32854)* 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195 = -427595 E(Total Points for Division 2 with 15 periods) = 10787*3 + (- 32854)* 12 = -361887 E(Total Points for Division 3 with 15 periods) = 10787*3 + (- 32854)* 12 = -361887 E(Total Points for Division 4 with 14 periods) = 10787*3 + (- 32854)* 11 = -329033 E(Total Points for all 4 divisions) = - 1480402 Condition 5, Condition 6, and Condition 7 Let the points awarded in period t of division k for performance be Ptk - Each period, 2 0 0 0 points were awarded (deducted) for every percent of achievement above (below) 80% of planned ROItk. Zero point was awarded if ROItk was 80% of planned ROItk. Planned ROItk = Optimal ROItk = 38.70% = 39% E(Ptk ) = E{[(ROItk /39) - 0.80]*100*2000} = 5128.2051 *E(ROItk ) - 160000 For t = 1, 2, and 3 , 7ttk = 8 ln(51) + 26 ln(101 - 50) = 133.68 since R&D& and P&Etk were fixed at $50 million each for the four periods prior to the first period. E(ROIik) = 33.68 (see Conditions 1 , 2, and 3 for calculations) Therefore, E(Plk) = 5128.2051*33.68 - 160000 = 12718 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196 E(Plk) = E(P2 k) = E(P 3 k) - 12718 For t > 4 , E ( 7 itk) = 124.48 (see Appendix D for calculations) E(ROl4 k) = 24.48 (see Conditions 1, 2, and 3 for calculations) E(P4 k ) = 5128.2051*24.48 - 160000 = -34462 Since E(P5 k ) to E(Pia stt, k ) are identical, for E(P5 k ) to E(Pia st t, k ): E(ROl5 k) = 24.48 (see Conditions 1, 2, and 3 for calculations) E(Psk) = 5128.2051*24.48 - 160000 - - 34462 E(Total Points for Division 1 with 17 periods) = 12718*3 + (- 34462)* 14 = -444314 E(Total Points for Division 2 with 15 periods) = 12718*3 + (- 34462)* 12 = -375390 E(Total Points for Division 3 with 15 periods) = 12718*3 + (- 34462)* 12 = -375390 E(Total Points for Division 4 with 14 periods) = 12718*3 + (- 34462)* 11 = -340928 E(Total Points for all 4 divisions) = - 1536022 Condition 8 Let the points awarded in period t of division k for performance be Ptk , the points awarded in period t of division k for ROItk performance be PROItk , the points Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 197 awarded in period t of division k for R&DPtk performance be PR&Dtk, and the points awarded in period t of division k for P&EOEtk performance be PP&Etk. Each period, 1850 points were awarded (deducted) for every percent of achievement above (below) 80% of planned ROItk. Zero point was awarded if ROItk was 80% of planned ROItk. 1850 points were awarded (deducted) for every % of achievement above (below) 80% of planned number of new products, R&DPtk. Zero point was awarded if R&DPtk was 80% of planned number. 550 points were awarded (deducted) for every % of achievement above (below) 80% of planned operating efficiency rating, P&EOEtk- Zero point was awarded if P&EOEtk was 80% of planned rating. Planned ROItk = Optimal ROItk = 38.70% = 39% E(PROItk) = E{[(ROV39) - 0.80]* 1 0 0 * 1850} = 4743.5897*E(ROItk ) - 148000 For t = 1,2, and 3 ,7 ttk = 8 ln(51) + 26 In (101 - 50) = 133.68 since R&Dtk and P&Etk were fixed at $50m each for the four periods prior to the first period. E(ROIik) - 33.68 (see Conditions 1 , 2 , and 3 for calculations) Therefore, E(PROIik ) = 4743.5897*33.68 - 148000 = 11764 Planned R&DPtk = Maximum R&DPtk = 1 0 ln(100 + 1) = 46 E(PR&Dtk ) = E{[(R&DPtk /46) - 0.80]*100*1850} = 4021.739 l*E(R&DPtk ) - 148000 E(R&DPik) = 36.61 (see Appendix D for calculations) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 198 Therefore, E(PR&Dlk) = 4021.7391*36.61 - 148000 - - 753 Planned P&EOEtk = Maximum P&EOEtk = 20 ln(100 + 1) = 92 E(PP&Etk ) = E{[(P&EOEtk /92) - 0.80]*100*550} = 597.8261 *E(P&EOEtk ) - 44000 E(P&EOE]k ) = 73.23 (see Appendix D for calculations) Therefore, E(PP&Eik ) = 597.8261*73.23 - 44000 = - 224 Since E(Pik ) = E(PROIik ) + E(PR&D]k ) + E(PP&Eik ), E(P]k)= 1 1 7 6 4 -7 5 3 -2 2 4 = 10787 E(Plk) = E(P2 k ) = E(P3 k ) = 10787 For t > 4, E (7 r tk) = 124.48 (see Appendix D for calculations) E(ROEk) - 24.48 (See Conditions 1,2,3 for calculations) E(PRORk ) = 4743.5897*24.48 - 148000 = -31877 E(PR&D4 k ) = -7 5 3 E(PP&E4 k ) = - 224 Therefore, E(P4 k ) = -3 1 8 7 7 -7 5 3 -2 2 4 = - 32854 Since E(PROIsk ) to E(PROIia s tt, k) were identical, for E(PROl5 k ) to E(PROIiastt, k): E(ROl5 k ) = 24.48 (see Conditions 1,2,3 for calculations) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. E(PROIsk) = 4743.5897*24.48 - 148000 = - 31877 E(PR&D5 k ) = -7 5 3 E(PP&E5 k ) = - 224 Therefore, E(P5 k) = - 31877 - 753 - 224 = - 32854 E(Total Points for Division 1 with 17 periods) = 10787*3 + (- 32854)*14 = -427595 E(Total Points for Division 2 with 15 periods) = 10787*3 + (- 32854)*12 = -361887 E(Total Points for Division 3 with 15 periods) = 10787*3 + (- 32854)* 12 = -361887 E(Total Points for Division 4 with 14 periods) = 10787*3 + (- 32854)*11 = -329033 E(Total Points for all 4 divisions) = - 1480402 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 200 Appendix F: Calculation of MSD between Actual and Optimal Investments Tangible Assets Firm (TAF) Optimal R&Dt = 23 Optimal P&Et - 77 Total number of periods = 17 periods in Division 1 + 15 periods in Division 2+15 periods in Division 3 + 14 periods in Division 4 = 61 MSD between actual R&Dt and optimal R&Dt = 1 (R&Dt - Optimal R&Dt) 2 /Total number of periods = H , I “ ' (R & D ,-23)2 /61 R&Dt = 100-P & E t Optimal R&Dt = 100 - Optimal P&Et MSD between actual P&Et and optimal P&Et = 4 (P&Et - Optimal P&Et) 2 /Total number of periods = y ; ; 1 [(100 - R&Dt) - (100 - Optimal R&Dt)] 2 /Total number of periods = y ; H y 1 ; ; ' ( - R&Dt + Optimal R&Dt) 2 /Total number of periods = MSD between actual R&Dt and optimal R&Dt Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 201 Intangible Assets Firm (ITAF) Optimal R&Dt = 77 Optimal P&Et = 23 Total number of periods = 61 MSD between actual R&Dt and optimal R&Dt = 1 (R&Dt - 77) 2 /Total number of periods R&Dt = 100- P&Et Optimal R&Dt = 100- Optimal P&Et MSD between actual P&Et and optimal P&Et = Xt=i * - Optimal P&Et) 2 /Total number of periods - ^ = 1 1 [(100 - R&Dt) - (100 - Optimal R&Dt)] 2 /Total number of periods = ' (- R&Dt + Optimal R&Dt) 2 /Total number of periods = MSD between actual R&Dt and optimal R&Dt Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Customer satisfaction, customer bargaining power, and financial performance

PDF

Corporate ethical orientation and crisis management before and after September 11, 2001

PDF

Board involvement in monitoring and strategy making: Antecedents and consequences

PDF

Feeling good about control: Design considerations for accountability systems in schools

PDF

Demographic diversity, team process, and team performance: Assessing moderator effects of cognitive conflict management practices and task interdependence

PDF

Essays on technological evolution and financial returns to innovation

PDF

Beyond skill development: The effects of training and development on the attitudes and retention of employees

PDF

Essays in supply chain management

PDF

Effects of redundancy in media coverage on nonprofessional investors' earnings forecasts

PDF

Androgyny and transformational leadership: Effects of gender and sex -role identity in the collectivistic context of Taiwan, R.O.C.

PDF

Influencing individual innovation through technology features that support cross -departmental understanding

PDF

A typology of corporate environemental strategy and its driving factors in jultinational corporations

PDF

Experimental evaluation of a new intervention designed to promote dementia caregivers' acceptance and empathic responding towards care -recipients

PDF

"Pay no attention to those men behind the curtains": An ethical examination of Los Angeles charter reform activities 1996--1999 by use of crisis management

PDF

Childhood videotaped neuromotor and social precursors of schizophrenia: A prospective investigation

PDF

Coevolution of technology and organization: A case study of BrokerHouse

PDF

A search for theory: Performance management to improve transportation safety

PDF

An evaluation of a brief HIV /AIDS prevention intervention using normative feedback to promote risk reduction among sexually active college students

PDF

Information systems capabilities and indirect relational value: Sustaining networked organizations

PDF

Defining the advice -seeking request in electronic solicitations in boundary spanning environments

Asset Metadata

Creator Ong, Lay Khim (author)

Core Title Does adding nonfinancial value drivers to a summary financial measure improve the learning and performance of managers?

School Graduate School

Degree Doctor of Philosophy

Degree Program Business Administration

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag business administration, accounting,business administration, management,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Merchant, Kenneth A. (committee chair), Bonner, Sarah (committee member), Earleywine, Mitchell (committee member), Van der Stede, Wim (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-375823

Unique identifier UC11335065

Identifier 3103953.pdf (filename),usctheses-c16-375823 (legacy record id)

Legacy Identifier 3103953.pdf

Dmrecord 375823

Document Type Dissertation

Rights Ong, Lay Khim

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA