THE ACCOUNTING REVIEW
V ol.78,No.3
2003
pp.725–758
Subjectivity and the Weighting of Performance Measures:Evidence from a
卫生标语
Balanced Scorecard
Christopher D.Ittner
贾诩David F.Larcker
Marshall W.Meyer
University of Pennsylvania
ABSTRACT:This study examines how different types of performance measures were
weighted in a subjective balanced scorecard bonus plan adopted by a majorfinancial
rvicesfirm.Drawing upon economic and psychological studies on performance eval-
uation and compensation criteria,we develop hypothes regarding the weights placed
on different types of measures.Wefind that the subjectivity in the scorecard plan
allowed superiors to reduce the‘‘balance’’in bonus awards by placing most of the
weight onfinancial measures,to incorporate factors other than the scorecard measures
in performance evaluations,to change evaluation criteria from quarter to quarter,to
1024成人ignore measures that were predictive of futurefinancial performance,and to weight
measures that were not predictive of desired results.This evidence suggests that
psychology-bad explanations may be equally or more relevant than economics-
bad explanations in explaining thefirm’s measurement practices.The high level of
subjectivity in the balanced scorecard plan led many branch managers to complain
about favoritism in bonus awards and uncertainty in the criteria being ud to determine
rewards.The system ultimately was abandoned in favor of a formulaic bonus plan
bad solely on revenues.
Keywords:balanced scorecard;subjective performance measures;nonfinancial per-formance measurement.
Data Availability:All data are proprietary.Confidentiality agreements prevent the au-
thors from distributing the data.
I.INTRODUCTION
T his study investigates the u of subjectivity in reward systems containing multiple performance measures.Bonus bad solely on profits and otherfinancial account-ing numbers have been criticized for encouraging managers to sacrifice long-run
We appreciate the comments of the reviewer and minar participants at the University of Alberta,University of Cambridge,Citibank Behavioral Science Rearch Council Conference,Georgia State University,University of New South Wales,University of Saskatchewan,Stanford University Summer Accounting Rearch Camp,The University of Texas at Austin,The University of Texas at Dallas,and University of Waterloo.This rearch was funded by the Citibank Behavioral Science Rearch Council,who support is gratefully acknowledged. Editor’s note:This paper was accepted by Terry Shevlin,Senior Editor.
Submitted September2002
Accepted February2003
725
726Ittner,Larcker,and Meyer performance to increa short-termfinancial results,and thereby maximize their bonus. To overcome the short-run orientation of accounting-bad reward systems,manyfirms are implementing compensation plans that supplementfinancial metrics with additional mea-sures in order to asss performance dimensions that are not captured in short-termfinancial results.The additional measures can take a variety of forms,ranging from quantitative,
nonfinancial metrics,such as employee and customer survey results,to qualitative asss-ments of performance by the manager’s superior.
One critical implementation issue that aris when incorporating multiple performance measures in reward systems is determining the relative weights to place on the various measures when determining bonus.One option is to u a formula that explicitly weights each measure.Potential difficulties with this option include determining the appropriate weights to place on each measure,the‘‘game-playing’’associated with any formula-bad plan,the possibility that bonus will be paid even when performance is‘‘unbalanced’’(i.e.,overachievement on some objectives and underachievement on others),and the like-lihood that all relevant dimensions of managerial performance are not captured by the lected performance ,Holmstrom and Milgrom1991;Baker et al.1994; Kaplan and Norton1996).
A cond option is to introduce subjectivity into the bonus award process.This sub-jectivity can take the form offlexibility in weighting quantitative performance measures when computing a manager’s bonus,the u of qualitative performance evaluations,and/ or the discretion to adjust bonus awards bad on factors other than the measures specified in the bonus contract.Some theoretical work indicates that greater subjectivity can improve incentive contracting becau it allows thefirm to expl
oit noncontractible information that might otherwi be ignored in formula-bad contracts,and to mitigate distortions in man-agerial effort by‘‘backing out’’dysfunctional behavior induced by incomplete objective performance measures(Baker et al.1994;Baiman and Rajan1995).However,other re-arch suggests that subjectivity in reward systems can lower managers’motivation by allowing evaluators to ignore certain types of performance measures that are included in the bonus plan,permitting bonus payout criteria to change each period,and introducing favoritism and bias into the reward ,Prendergast and Topel1993).As a result, managers will be less able to distinguish what constitutes good performance,less likely to believe that rewards are contingent on performance,and less convinced that performance criteria are being applied consistently across the organization.
Drawing upon economic and psychological studies on the choice of performance mea-sures for performance evaluation and compensation purpos,we develop exploratory hy-pothes regarding the weights placed on different types of performance ,financial versus nonfinancial,quantitative versus qualitative,and input versus outcome)in subjective bonus computations.We test the hypothes using quantitative and qualitative data gathered during an extensive,multiyearfield investigation of a balanced scorecard bonus plan in the ail banking operations of Global Financial Services(GFS),a leading internationalfinancial rvices provider.怎么去新加坡
In1993,GFS replaced a profit-bad bonus plan for branch managers with a formula-bad system that rewarded multiple accounting and growth measures once customer sat-isfaction and operational audit hurdles were achieved.This system changed rapidly during the nine quarters it was in u as the bank sought to eliminate gaming and promote per-formance across a broader t of measures.The formula-bad plan was replaced in the cond quarter of1995by a‘‘balanced scorecard’’bonus plan containing six categories of financial and nonfinancial performance measures,some of which were bad on qualitative evaluations by the managers’supervisors.Unlike the earlier formula-bad plan,subjective The Accounting Review,July2003
Subjectivity and the Weighting of Performance Measures 727The Accounting Review,July 2003
weightings were ud to aggregate the various scorecard measures when determining overall performance evaluations and bonus.
Although Kaplan and Norton (2001)cite the GFS bonus plan as an example of a scorecard-bad reward system that prevented managers from underperforming on any of the scorecard dimensions,we find that the u of subjectivity in weighting the scorecard measures allowed supervisors to ignore many performance measures,even though some of the measures were leadi
ng indicators of the bank’s two strategic objectives (financial performance and growth in customers).Instead,short-term financial performance measures become the primary determinant of bonus.In addition,a large proportion of branch managers’performance evaluations was bad on factors other than the scorecard measures,even though discretion to consider other factors was not a component of the bonus plan.The move from the formula-bad system to the more subjective scorecard led many branch managers to complain about favoritism in bonus awards and uncertainty in the criteria being ud to determine rewards,and caud corporate executives and human resource managers to question the scorecard’s u for compensation purpos.Ultimately,the company aban-doned the scorecard plan at the end of 1998in favor of a commission-style system bad on revenues.
Our study makes four contributions to the performance evaluation and compensation literatures.First,we extend cross-ctional studies on the u of subjectivity and discretion in bonus plans (e.g.,Murphy and Oyer 2001;Gibbs et al.2002).Whereas the studies focus on the factors explaining who includes subjectivity in compensation contracts,our study emphasizes how subjectivity is incorporated into performance evaluations and bonus awards.Second,we provide further evidence on the influence of informativeness on per-formance measure weighting.Prior studie
s on the relative weights placed on financial and nonfinancial performance measures (e.g.,Bushman et al.1996;Ittner et al.1997)generally include proxies for the noi in financial measures,but do not include direct measures of the informativeness of nonfinancial measures due to data constraints.The detailed data in our study allow us to develop stronger tests of economic theories that the relative weights placed on performance measures other than financial results are a function of their inform-ativeness.Third,our rearch complements psychology-bad experimental work on the importance placed on various types of performance measures by examining whether their experimental results hold in an actual performance-evaluation tting.Finally,although the ability to generalize our results is limited by the analysis of only a single firm,we provide one of the first detailed studies of scorecard-bad compensation plans.1Despite survey evidence that a growing number of firms are using balanced scorecards for compensation purpos (Kaplan and Norton 2001,Chapter 10),relatively little is known about the im-plementation issues associated with scorecard-bad reward systems.
The remainder of the paper is organized in five ctions.Section II reviews related rearch on subjective performance evaluation and performance measure weighting,and develops our exploratory hypothes.The rearch tting for our study is described in Section III.The statistical te
sts for our rearch hypothes are prented in Section IV .In Section V we provide detailed qualitative analysis regarding the success of the balanced scorecard implementation.The final ction summarizes our rearch results and discuss implications for future rearch.
1See Malina and Selto (2001)and Campbell et al.(2002)for field-bad studies examining other us of the balanced scorecard.Also e Banker et al.(2000)for field evidence on the implementation of a compensation plan incorporating nonfinancial measures.
728Ittner,Larcker,and Meyer
The Accounting Review,July 2003
II.LITERATURE REVIEW AND HYPOTHESES
Performance Measure U in Subjective Performance Evaluation
Kaplan and Norton (1996,10)argue that balanced scorecards should reflect four types of measures:(1)financial and nonfinancial;(2)external (financial and customer)and in-ternal (critical business process,innovation,and learning and growth);(3)inputs/drivers and outcomes/results;and (4)objective,easily quantifiable measures and more subjective,judgmental measures.Although Kapla
n and Norton (1996,2001)provide little guidance on how to combine or ‘‘balance’’the disparate measures when evaluating managerial per-formance,they conjecture that the balanced scorecard renders subjective reward systems ‘‘easier and more defensible and also less susceptible to game playing.’’(See Kaplan and Norton 1996,220.)
Analytical rearch on the u of subjectivity in performance evaluation and compen-sation focus on the benefits of subjective bonus awards (e.g.,Baker et al.1994;Baiman and Rajan 1995),the drawbacks of subjective performance evaluations (e.g.,Prendergast and Topel 1996;MacLeod 2001),and the factors influencing the relative weights placed on subjective versus objective performance measures in incentive contracts (e.g.,Murphy and Oyer 2001).Most of the models do not examine how different types of performance measures or different forms of subjectivity (i.e.,flexibility in assigning weights to measures,u of qualitative performance evaluations,and/or discretion to incorporate other perform-ance criteria)should be incorporated into subjective bonus awards.An exception is Murphy and Oyer’s (2001)model,which suggests that the relative weight on subjective measures will be higher in privately held companies,larger companies with more top managers,less autonomous business units,companies with substantial growth opportunities,and companies where accounting profits and shareholder returns are less highly correlated.Their cross-ctional empirical tests of executive bonus provide mixed support for the hypothes.
A related empirical study of automobile dealerships by Gibbs et al.(2002)finds that subjectivity (defined as the prence of any subjective bonus payout or as ‘‘discretionary bonus’’as a percent of total compensation)is positively related to departmental inter-dependencies,financial loss,the manager’s tenure,and the achievability of formula-bad bonus.While the two empirical studies provide insight into who us subjectivity in compensation contracts,they provide little insight into how subjectivity is applied or per-formance is evaluated when multiple types of performance measures are incorporated into the bonus contract.2
dota2地图Other studies address the relative importance placed on the various types of measures highlighted by Kaplan and Norton (1996).The studies fall into two rearch streams.The first stream focus on economic models of incentive contracting.Economics-bad agency models emphasize the role of performance measure choice in aligning agents’goals with tho of the principal,and indicate that the choice of performance measures in incentive contracts should be a function of the informativeness (or incremental information content)of each measure regarding the worker’s action choices (e.g.,Holmstrom 1979;Banker and Datar 1989;Feltham and Xie 1994;Hemmer 1996;Lambert 2001).
下一步工作A cond rearch stream adopts a psychological perspective.The studies examine how human in
formation-processing capabilities and decision strategies influence the types of information individuals u when asssing performance.The behavioral experiments 2Murphy and Oyer’s (2001)tests examine the u of individual performance appraisals and discretionary allo-cation of bonus awards,but do not examine the u of specific types of performance measures.Gibbs et al.(2002)u an aggregate measure of subjectivity,which their survey defines as bonus awarded bad on a supervisor’s subjective judgment of a manager’s performance.
Subjectivity and the Weighting of Performance Measures729 suggest that issues such as information overload and cognitive bias can play a significant role in the relative weights placed on different types of balanced scorecard , Lipe and Salterio2000,2002).In particular,this rearchfinds that evaluators frequently place greater or exclusive emphasis on certain types of measures,even when other types of measures also provide relevant information on the subordinate’s performance.泰和县
Economics-Bad Hypothes
Informativeness
Economics-bad agency models focus on the role of performance measures in pro-moting congrue
nce between the principal’s objective and that of the ,Holmstrom 1979;Banker and Datar1989;Lambert2001).Two primary insights from the models are that compensation contracts should include any(costless)measure that carries incremental information on the agent’s actions,and that the relative weight placed on an individual measure should be a function of the measure’s signal-to-noi ratio,as reflected in its nsitivity(or the change in its mean value in respon to a change in the agent’s action) and precision(or the inver of the variance in the measure given the agent’s action).
Although many of the models say little about the specific types of performance mea-sures that should be included in compensation contracts,veral studies extend the papers to investigate the role of nonfinancial measures(Feltham and Xie1994;Haur et al.1994; Hemmer1996).The models suggest thatfinancial measures alone are unlikely to be the most efficient means to motivate employees,and demonstrate how incentives bad on nonfinancial measures can improve contracting by incorporating information on agents’actions that is not fully captured in contemporaneousfinancial results.
A number of cross-ctional empirical studies draw upon informativeness theories when examining the relative weights placed on individual,nonfinancial,or subjective performance ,Bush盐渎
man et al.1996;Ittner et al.1997;Murphy and Oyer2001;Ittner and Larcker2002).The studies typically u two approaches to test the models’predictions. First,the effects of noi on performance measure choice are examined using the variance in objective,financial measures.In the tests,the weight placed onfinancial measures is predicted to decrea in their noisiness,while the weight on other types of measures is predicted to increa.Second,proxies for factors that are expected to influence the inform-ativeness of individual,nonfinancial,or subjective ,growth opportunities, strategy,product life cycle,etc.)are examined,with the weight placed on the measures expected to be higher when the informativeness proxies are greater.
The results from the studies are mixed.While proxies for the factors predicted to influence informativeness are generally associated with incread weights on the mea-sures,proxies for the noi infinancial measures tend to have little association with measurement choices.However,a significant limitation of the analys is the lack of data on nonfinancial or subjective performance dimensions,forcing rearchers to u indirect proxies for the measures’informativeness.
Consistent with the nonfinancial performance measure and balanced scorecard litera-tures,we assume that currentfinancial measures are potentially incomplete,and that other indicators of futurefinancial performance can provide incremental information on the man-ager’s ,Felth
am and Xie1994;Hemmer1996;Kaplan and Norton1996,2001). We also assume that measures that are more predictive of future performance provide greater information on the congruence between the agent’s actions and the outcomes desired by the principal.In an agency tting,the coefficients(or weights)associated with the nonfinancial performance measures in the structural model linking nonfinancial measures to futurefinancial ,the‘‘business model’’)and the coefficients(or weights)ud
The Accounting Review,July2003
