Quản trị kinh doanh - Chapter 9: Performance evaluation

Chapter 9: Performance Evaluation4/8/2013Chapter 9: Performance EvaluationPurposes of Performance EvaluationWays to Evaluate PerformanceQuantitative Performance MeasurementRisk ProfileDistortionCommon Distortions in Performance MeasuresIntangibles Opportunity costsGroup sizeTime horizonManipulationMatch of the Performance Measure to Job DesignSubjective EvaluationWhy Use Subjective Evaluations?Improve Consideration of Risk in the EvaluationReduce Distortion of ManipulationImprove Incentives for Risk TakingImprove Decision MakingGive the Incentive System More FlexibilityExpand CommunicationImprove TrainingPractical ConsiderationsWho Should Evaluation Whom? Fairness, Bias, and Influence CostsGiving and Receiving Subjective EvaluationsDifferent Roles of EvaluationHow Frequently Should Evaluations Be Made?Summary 4/8/2013Chapter 9 – Performance EvaluationAfter Today’s Lecture, you will be able to address the following questions:Should a firm use only quantifiable metrics to support employee evaluations?Are subjective measures ever appropriate?What are (un)controllable events in this context and why is the concept important? How much information should a firm include in its evaluations?Is ranking a useful form of evaluation?4/8/2013Performance Evaluation DesignObjective is to fully capture all of the employee’s effect on firm valueWhen such a measure can be established, further measurement is not advisableEvaluations can include either subjective, quantitative and/or qualitative measuresFive dimensions for designing measures:Risk; Distortion; Scope; Manipulation; Job Design 4/8/20132. Quantitative Performance Measures (QPMs)Quantifying performance is clearly preferred to subjective evaluation if possibleStart by choosing one or more performance measuresWhat should you look for?General properties of QPMsrisk profiledistortionpotential for manipulation4/8/2013Risk ProfileIt is important to distinguish 2 kinds of “risk”Uncontrollable Risk = random events employee cannot foresee or react to“classical” concept of riskex: macroeconomic conditionsControllable Risk = (possibly random) events employee can, to some extent, foresee or react totypically, employee “knows more than you” b/c they do the jobex: customer requests in store that dayBoth affect the performance measure, but they have different implications for incentives4/8/2013Uncontrollable RiskExposing an employee to more uncontrollable risk raises costs of the incentive plane.g., suppose Pay = a + b•(Perf. Measure) = a + b•(True Perf. + )a = base salary; b = commission rate is uncontrollable riskif variance in  = ², variance in pay = b²² stronger incentives b means greater risk for the employeeHow can we address this?spend to measure performance more accurately (filter more risk)raise base salary a to provide more risk premiumlower commission b to reduce risk but also lower incentives4/8/2013Controllable RiskWe may not want to filter out controllable riskwant to motivate employee to respond to it, so we may want to include this kind of risk in the performance measureuse information that arises as they do their jobforesee, prevent, react to events where possibletake initiativeIf the job entails more controllable riskuse a measure that includes its effects to the extent possibleoften this is what we’ll call a “broader” measuregive stronger incentives, not weaker4/8/2013Risk in MeasureIncluding measures that are even partially uncontrollable introduces risk in evaluation for the employeeThe more controllable a measure, the lower the risk to the employeeIncentive intensity can be higher with more controllable measures4/8/2013ScopeYou often have to choose the scope of a measurenarrow measures focus on fewer aspects of performance; broad on moreexample: department v. firm profitNarrow Broad(fewer factors) (more factors)High Distortion Low DistortionLow Risk High Risk4/8/2013A useful way to think about these is the Scope of a measurenarrow measures focus on fewer aspects of performance; broad on moreex: department v. firm profitYou face a key tradeoffwe might use a narrow measure only because it reduces riskbroad measures: more controllable risk, distort less, harder to manipulateuse broader measures in less specialized jobs & w/ more decentralization4/8/2013Scope of MeasureBroader evaluation designs that include more measures may incorporate more riskNarrow, focused measures can reduce riskNarrow measures can introduce performance distortions owing to undue focusN.B., risk exerts disproportionate influenceImportant to match measures to job design and if necessary, design job to available measures4/8/2013DistortionAlways remember: “You get what you pay for”any measure other than “firm value” distorts incentivesex any measuredistortions tend to arise 2 waysfocus on what is easy to measuremeasure doesn’t match job designThe more distorted the measure, the more weakly should you tie it to rewards4/8/2013Scope: Trading Off Risk & DistortionHow many tasks to include?revenue v. costhow does a job’s specialization / multitasking affect perf. measure scope?Include proxies for intangibles?quality, customer satisfaction, opportunity costs (EVA)How large a work unit should be measured?individual, team, unit, division, firmWhat time horizon should be used?most metrics are backward rather than forward looking4/8/2013Additional ConsiderationsSources of distortion include:IntangiblesOpportunity costsTimingAdditional considerations for subjective evaluationsAvoid hindsight biasUse prospectivelyAdd richness to feedback and improve quantitative measures 4/8/2013Risk vs. ScopeOptimal evaluation design balances the risks of added measures against the distortions potentially resulting from limited measuresPoorly designed evaluation systems will be “gamed” by employees, to firm detrimentEvaluations (and job designs) should be updated and adapted as needed to match the current organizational setting*4/8/2013Performance EvaluationFirst, choose a quantitative measure matching job as closely as possibleNext, askhow much is the measure affected by uncontrollable risk? if a lot, consider narrower scoope, or RPE if risks are similar across employees you’ll have to pay more (risk premium) for more riskor is risk controllable? in a dynamic environment employee can control, do opposite (& decentralize)what am I missing? How will it distort employee’s behavior? (it will) if a lot, consider broadening the focus (or adding a second measure)how could the employee manipulate the evaluation?The more flawed the measure, the weaker should the incentive beFinally, use subjective evaluation / discretionary bonuses to “fix” problems & make overall incentives better4/8/2013Example: TopGradingPioneered at GE, tried by many other firmssome form of forced curveE.g.: top 20%; middle 70%; bottom 10%those w/ consistently low ratings put on “Performance Improvement Plan” improve, or re-deployed w/in GE, or forced outWhere does it work best?larger workgroupsteamwork is less importantculture is more aggressive / Darwinianextensive attention is paid to coaching & developmentemployees are given clear expectations about weeding outpotential legal liabilities are monitored carefullycare is taken to make sure evaluations are based on performance4/8/2013Potential for ManipulationAdditional problem: employee may manipulate performancetake actions that improve the measure, but not firm valueex: Colorado potato beetles; virtually any accounting #The more manipulable the measure, the less should it be tied to rewardsnote – measures tend to degrade over time: the longer it has been used, the more has the employee figured out how to manipulatenarrower measures (below) are usually easier to manipulate4/8/2013Match to Job DesignNarrow measures are used to reduce uncontrollable riskIf there is much controllable risk, consider a broader measurecapture broader impact of many variables the employee can react tosimilarly, decentralize moreput another way, an ideal measure matches the responsibility of the jobexample: Table 1 (next slide)how does more decentralization affect a job’s perf. evaluation scope?4/8/2013Set Overall Level of PaySet to achieve desired recruitment & retention of talentcompensation surveysrates of applicants, offer acceptance & turnoverPay risk premium to compensate for incentives on risky evaluationsPay effort premium to compensate for job with strong incentives4/8/2013Maximizing Employee PerformanceThree ways of looking at the questionefficiency  specialization & Classical job designlearning  specific knowledge, Modern job design, trainingmotivation  Modern job design, pay for performanceClassical design isn’t as important these days, butnever forget the important benefits of specialization, centralizationdon’t get too enthusiastic about management fads; remember the important historical ideas & lessonsIn any job these days, use elements of the latter two, balanced against Classical elements 4/8/2013Maximizing Employee PerformanceModern job design & pay for performance maximize employee potential in several powerful waysdesign jobs to maximize employee learningfor many workers, this also drives powerful intrinsic motivationprovide training in problem-solving skills to reinforce the learningprovide evaluations that fairly & reasonably match responsibility, then “set the employee loose” to maximize performancereward that performance to drive “intrapreneurship”In other words, provide the opportunity, method & motivation for high performanceuse the principles discussed to better design jobs, evaluate, & rewardIf you’ve done this well, step back & let them perform!accepting some loss of your control, for the sake of creativity4/8/2013Subjective EvaluationAlternative to numeric evaluationTypical problemsinflation & compression of ratingsreluctance for negative feedbackde-motivating effects?favoritism, “yes men,” pressure on evaluatorweak evaluator incentiveslow trust of evaluatorhalf-life of appraisal systems  5 yrsEveryone hates subjective evaluation, so why is it always used?4/8/2013Subjective EvaluationSubjective evaluations can be made with precisionUsed singularly, as a complement or as a filter for quantified measuresManager’s role and level of effort can be significantExtremely useful for reducing measurement risk – applying judgment to the unanticipated 4/8/2013Why Use Subjective Evaluations?Good evaluation requires judgment, because metrics are imperfectSubjectivity, if done well, improves incentives many waysreduce uncontrollable risk, distortion, & manipulationwhat is “uncontrollable” & what is “controllable”?improve risk takingimprove decision making (use of controllable risk)encourage pro-active preparation, effective reactiongive incentive system flexibilityexpand communication between manager & employeebe a form of training4/8/2013Conducting a Subjective EvaluationFirst, think retrospectively: “what did he know & when did he know it?”avoid hindsight bias: given what was known, were correct decisions made?if not foreseeable, were responses appropriate?did employee try to “pass the buck” or take responsibility? reward performanceMore importantly, think prospectively: what can be improved?information, training, etc. responsibilities, future evaluation criteria & expectationsGoal: performance management, not “drive-by evaluation”4/8/2013Improving Subjective EvaluationMultiple evaluators (if feasible) may reduce bias & favoritismProvide supervisors w/ performance data, training in evaluation, structured forms to give some guidanceProvide oversight & allow employees to challenge evaluationsCulture of constructive feedback giving & receivingLeadership must make careful & fair evaluation a high priorityevaluate on ability to do performance management 4/8/2013Process, not Just PolicyGood incentives require a fair systemability to modify & evolve the planexperimentation, communication (in both directions)subjective evaluationspay attention to training, monitoring & motivation of evaluatorsMBO-style approach turns evaluation into on-going performance management Constant tension between coaching / development & evaluation4/8/2013Tie Pay to PerformanceGive stronger incentives ifsorting for talent is more importanteffort is more costly to employeeseffort has greater impact on firm valueemployee has valuable specific knowledgeintrinsic motivation is relatively low or distorts behaviorevaluation is more effectiveless risky, distorted, or manipulablewell balanced across multiple taskssubjective evaluations are trusted more4/8/2013Recommendations[UK Financial Services Authority Report, 2012]Reverse Engineeringhow will their behavior be distorted?how will employees manipulate the plan?how will they (mis)manage risks? Try to pass blame?Scale up the intensity of monitoring with risksConsider “clawbacks” or bonus banks; bonus capsDon’t forget supervisor incentivesemphasize quality more than quantity4/8/2013Receiving an AppraisalAsk for feedback, regularlyAvoid defensivenesstake responsibility for mistakes; don’t politicize / lobbyFocus on constructive suggestionsset goals; discuss prioritiesIdentify & ask for what you neednew skills; information; resources4/8/2013RECAP: Chapter 9 – Performance EvaluationThe ideal employee evaluation measures the contribution to firm valueThe firm can apply quantitative and/or qualitative measures considering risk and scope in their designSubjective evaluations can be used to replace or complement available metrics – reducing risk and distortionUncontrollable events are those whose effect the employee could not anticipate or mitigateTruly uncontrollable events should not be included in the evaluationMore is not necessarily better. In general,More measures => more riskLess measures => greater distortion4/8/2013Part III: Paying for Performance4/8/2013Part Three – Paying for PerformanceNew Concepts IntroducedMoral HazardPrincipal-Agent relationshipMarginal IncentiveTournament Theory *4/8/2013Moral Hazard[Definition also found in your textbooks]Tendency to take undue risks when the consequences are not fully borne by the party taking the actionPaul Krugman: “Any situation in which one person makes the decision about how much risk to take, while someone else bears the costs if things go badly” Examples? 4/8/2013Principal-Agent ProblemIs a subset of the MORAL HAZARD PROBLEMOccurs when one individual or entity is called upon to act on behalf of anotherCalled Upon can include a contact relationship or employmentTypically, the Principal is the employer and the Agent is the employee – but this is not universally the caseIn economic terms, the P-A Problem involves INFORMATION ASSYMETRYUNCERTAINTY OF MEASUREMENT AND OUTCOME RISKIs addressed by incorporation of RISK into agent’s compensation to the degree practicable4/8/2013Marginal IncentiveAs introduced in Lazear & Gibbs Ch9 Marginal incentive for an employee must consider the cause-effect relationship betweenIncremental effortChange in performance measureResponse of compensation systemMarginal benefit to the company 4/8/2013Tournament Theory of Executive CompensationPossibly the most insightful result from Personnel EconomicsElegant description of the real worldStates that paying an individual more than his/her marginal product can be justified if those beneath are motivated by the prizeWorks well when relative performance assessment is available rather than absolute Profound implications for organizational design!4/8/2013Summary: Intro to Pay for PerformancePrincipal-Agent Problem is fundamental to personnelAddressed by using measures closest to output and aligned with principal’s interestsTournament Theory is widely relevant to real world settingsBalance certainty of measure and need for team cooperation against high rewardMeasurement and Evaluation translate firm objectives to employee reward and feedbackReview job expectations in terms of what is controllable and what is not controllableBalance quantitative and qualitative measuresAdd fidelity to assessments by limiting noise – scope and variability4/8/2013