Effectiveness of financial incentives as motivators for sales personnelMoney and employee motivation Abstract. – Research consistently substantiates the effectiveness of financial incentiveson job performance, although companies need to consider the issue of job quantity versusquality and also be aware of the limitations of financial incentives. Employees can havevastly different motives for acquiring wealth – including using money to fulfillpsychological needs. Thus, it is not surprising that money alone is less an effectivemotivator for employees than when it is used in conjunction with non-financialreinforcements. We review the nuances of financial incentives and make basicrecommendations that can form the basis of best practice compensation and incentivepolicies.The 20_20 Skills™ assessment is the only screening, selection and training assessmentguaranteed to satisfy professional testing and legal standards and which was designedspecifically for the hospitality-service industry. During the assessment test takers mustindicate the degree to which they disagree or agree with a number of statements thatpertain to their personality or approach to professional situations. One of those statementsis:“Acceptance is more important to me than money”How well does that statement describe you, your coworkers or executive team? Moreimportantly, what does one’s attitude towards money as a motivator actually reveal? Thisarticle explores this latter question in the context of classic and recent research findingspertaining to the workplace and beyond. Our aim is to outline the nuances of financialincentives for enhancing job performance.Understanding Materialism

Materialism is defined simply as when a person values money, wealth and possessionsover other things in life14. Studies have consistently shown that a materialistic focus inlife is associated with a lower psychological well-being2,6,7,8,12. Even though individualswho are very poor financially demonstrate increased happiness when their income rises,intensity of desire for wealth remains negatively correlated with psychological wellbeing2.There are a few possible reasons for these trends2. Specifically, materialisticpursuits do not provide what people are really looking for in their quest for happiness; thetendency for those without much in the way of non-material resources to focus onmaterial goods or the endless and forever evolving supply of goods and servicesproduced in materialistic or capitalistic societies precludes satiation. They furtherMoney and employee motivation 3.suggested that people who are unhappy or lacking in social connections may seek solacein material goods, using external means to fulfill internal desires and aspirations2.Some researchers16 have argued that the motives that drive one to focus on money aremore relevant than the intensity of the focus itself, and they subsequently differentiatebetween instrumental materialism and terminal materialism. Instrumental materialismdescribes using material goods as a means for attaining personal goals and fulfillment;whereas terminal materialism involves using material possessions to achieve social statusand elicit envy from others. They also divide motives for materialistic pursuit into threedifferent categories: positive, negative and freedom of action. Positive motives involveusing money for basic necessities and as a measure of achievement. Negative motivesrefer to using money to gain power or superiority over others. Negative motives alsoinclude efforts to allay one’s self-doubt. Motives concerning freedom of action simplyimply spending money in any way that one desires.When the significant negative correlation between subjective well-being and money

importance was analyzed, while controlling for the influence of motivation, thecorrelation lost its statistical significance. In addition, when calculated independently,the relationship between negative motives and money importance was both negative andstatistically significant. As a result, the authors16 attributed the correlation betweenpsychological well-being and money importance to negative motives alone. This suggeststhat not all motives for wanting money lead to decreased happiness. Their findingssuggest that it is not the importance of money that contributes to well-being; rather,attempting to use money to alleviate self-doubt and increase self-esteem leads toproblems. They add that such motives become problematic when money is used forthings it cannot provide, e.g., self-esteem, happiness and genuine friendship.Overall, the research described above suggests that having money is not related topersonal happiness per se. However, whether people who have money or not, the amountof importance they place on it can actually be maladaptive. Indeed, individuals whoconsider money important tend to be less happy than those who place less importance onmoney2,16.

Financial Incentives in the WorkplaceNo one works for free, nor should they. While pursuing money based on negative motives can lead to a poorer psychological well-being, this is not the same as pursuingmoney to provide security and comfort for oneself and family. Obviously, employees want to earn fair wages and salaries, and employers want their workers to feel that iswhat they are getting. To that end, it is logical that employees and employers alike view money as the fundamental incentive for satisfactory job performance.The use of monetary or other financial incentives in the classic “work performance paradigm” is based primarily on reinforcement theory. Reinforcement theory15 focuses

on the relationship between a target behavior (e.g., work performance) and its consequences Money and employee motivation 4. (e.g., pay), and it is premised on the principles and techniques of organizational behaviormodification9,17. Organizational behavior modification is a framework within which employee behaviors are identified, measured and analyzed in terms of their functional consequences (i.e., existing reinforcements) and where an intervention is developed using principles of reinforcement10,17. In a much publicized study, Gupta and her colleagues5 analyzed thirty-nine studiesconducted over four decades and found that cold-hard cash motivates workers whether their jobs are exciting or mundane, in labs and real-world settings alike. But the research team acknowledges that money is not the only thing that concerns employees – noting that beyond a certain point higher salaries will make employees happier, but it will not “buy” better performance. Still, Gupta warns that employers who dole out small merit raises – less than 7% of base pay – may do more harm than good. According to her, small raises can actually be dysfunctional in terms of motivation because employees become irritated that their hard work yielded so little. Because of this, she advises employers who must give small raises to be careful about linking them to results and to be scrupulousabout being fair. Still, the research by Gupta’s team is just one study from a wealth of findings that can appear inconsistent. These apparent inconsistencies reflect, in part, important nuances about the relationship between monetary incentives and job performance. For example,the broad literature on job performance encompasses financial incentives that address both individual and group performance and productivity. Furthermore, monetary

incentives can extend beyond the mere raises discussed by Gupta’s team to include individual and small-group rewards, merit pay, pay-for-performance, variable pay plans, or group bonus plans as well as profit-sharing and gain-sharing incentive plans. Perry and his colleagues13 analyzed this diverse literature in an outstanding review and culled two general propositions relevant to best practices in the service industry:Financial incentives moderately to significantly improve task performance,but their effectiveness is dependent upon organizational conditions.Differences in institutional arrangements contribute to the feasibility and effectiveness of various monetary incentives, as do differences in employees’ preferences for specific incentives. Therefore, companies are wise to study these issues before implementing changes to existing incentive plans. This is especially pertinent for service organizations, where financial reinforcements tend toproduce a stronger effect on task performance than non financial rewards used alone. Even stronger results are seen with a composite approach. For example, one meta-analysis of 72 field studies found that monetary incentives improved task performance by 23%, social recognition improved task performance by 17% and feedback elicited a 10% improvement18. Simultaneously combining all three types of reinforcements improved performance by 45%.Group incentive systems are consistently effective in private sector settings.Team-based or small-group incentives are defined as rewards whereby a portionMoney and employee motivation 5.of individual pay is contingent on measurable group performance1. In general, its

effectiveness is dependent on the characteristics of the reward system, theorganization, the team and the individual team members1. Here again, studyingthis issue via employee surveys or interviews can be useful. But generallyspeaking, research suggests that equally divided small-group incentives sustainhigh levels of productivity and satisfaction for group members, and that smallgroupincentives are at least as effective as individual incentives with groups oftwo to twelve people4.

Qualitative, quantitative and survey research studies of alternative pay systemssuch as profit-sharing or gain-sharing plans are even more consistent in theirfindings. These incentive programs include various pay-for-performanceapproaches that link financial rewards for employees to improvements in theperformance of the work unit20. Research reveals that these types of incentivesystems are associated in practice – and in employer and employee minds – withboth higher productivity and improvements in organizational performance.Cashing in on the Latest ResearchWe have shared several useful insights into the relationship between money andemployee motivation. For readers wanting more in-depth discussions of this information,we recommend the resources listed below. These sources also provide guidance on howto translate some of these principles into practice:_ Heneman, R. L. (1992). Merit pay: linking pay increases to performance Ratings.New York: Addison-Wesley._ Milkovich, G. T., & Wigdor, A. K. (Eds). (1991). Pay for performance:evaluating performance appraisal and merit pay. Washington, DC: NationalAcademy Press._ Perry, J. L., Mesch, D., & Paarlberg, L. (2006). Motivating employees in a newgovernance era: The performance paradigm revisited. Public AdministrationReview, 66, 505-514. (Accessed on October 26, 2006at http://www.aspanet.org/scriptcontent/custom/staticcontent/t2pdownloads/PerryArticle.pdf)

Of course, the available research makes it clear that a monetary incentive is simply onepiece of a larger mosaic of issues. Savvy organizations delve into those other issues.Employee surveys and interviews can help organizations identify what type of monetaryincentives are the most motivating to a given employee base, and, naturally, thoseincentives must be feasible for the organization to implement. Subsequently, companiesmust balance financial considerations with other non-financial reinforcements tomaximize job quantity and quality. The best rule of thumb stems from Stajkovic andMoney and employee motivation 6.Luthans’ influential meta-analysis,18 which concluded that feedback combined withmoney and social recognition produced the strongest effect on job performance.We further recommend that performance feedback should be structured to maximize itsbenefit both to the employee and the company. For example, many employees respondwell to coaching or mentoring. And there are many tools, such as the 20_20 Skills™assessment, that can streamline goal-setting for individual employees or even entiredepartments. Goal-setting is an important aspect to enhancing job performance, althoughit is not always an obvious one to managers. For instance, one review19

discovered that,while monetary incentives influence performance, the relationship is not mediated bygoal-setting. That is, goal-setting and monetary incentives independently influence jobperformance (for more information, see Perry et al.13).HVS has also completed proprietary research on the effectiveness of pay-for-performancemetrics in the hotel, gaming and restaurant industries, and its results generally reinforcethe findings of researchers like Gupta and Perry. That is, “money ranks below otherfactors relative to job satisfaction, but high on the motivational matrix when all otherfactors have been met.” Furthermore, as executive search professionals, we understandwhat motivates individuals to change positions and employers. Job stature, scope of

responsibilities, employer reputation and other key criteria need to be met beforecompensation negotiations begin. Once negotiations begin, it is all about the money andhow the pay package compares to industry peers.For example, once major league baseball players become free agents, “what motivatesthem to stay or sign with another team?” Issues such as team dynamics, ability to win,job security, preferred location and position must be satisfied. Once satisfied, it becomesan issue of money. And the easiest way to compare offers is to look at what other similarplayers in similar positions are paid. It is no different in the corporate world. The tangibleresults of a well-conceived pay program are numerous, but the HVS study found thatexecutives with clear pay-for-performance metrics outperformed their peers by nearly80% at the Net Operating Income line.HVS Executive Search and 20_20 Skills™ assessment offer consultation on employeecompensation and coaching issues. But for those companies seeking simple guidancehere and now, the following heuristic nicely summarizes how most employees in theprivate sector view the relationship between money and employee motivation…“Show me the money, show me respect and show me attention…or show me the door.”Money and employee motivation 7.About the AuthorsJames Houran holds a Ph.D. in Psychology and is president of 20_20 Skills™Employee Assessment (www.2020skills.com). A full member of the AmericanPsychological Association and the American Psychological Society, Dr. Houran is a 15-year veteran in research and assessment on peak performance and experiences – with aspecial focus on online testing. He has authored more than 100 journal articles, and hisaward-winning work has been profiled by a myriad of media outlets and programsincluding the Discovery Channel, A&E, BBC, NBC’s Today show, Wilson Quarterly,USA Today, New Scientist, Psychology Today, Forbes.com and Rolling Stone.

Keith Kefgen holds a BS in Hotel Administration from Cornell University and ispresident of HVS Executive Search. HVS Executive Search is the leading executivesearch firm specializing in the gaming, lodging and restaurant industries with offices inNew York, Hong Kong, Las Vegas, London, Moscow and New Delhi. Mr. Kefgen is afrequent lecturer on industry related issues and has written more than 100 articles on thetopics of executive selection, pay-for-performance, corporate governance and executiveleadership. He serves on the board of the International Association of Corporate &Professional Recruitment (IACPR) and the Association of Executive Search Consultants(AESC).For information on the Best Practice 20_20 Skills™ assessment system, contact:James Houran, Ph.D.jhouran@2020skills.com516.248.8828 x 264www.2020skills.comFor information on Best Practice Executive Search, Corporate Governance andCompensation, contact:Keith Kefgenkkefgen@hvsinternational.com516.248.8828 x 220Money and employee motivation 8.References1DeMatteo, J. S., Eby, L. T., & Sundstrom, E. (1998). Team-based rewards: currentempirical evidence and directions for future research. Research in OrganizationalBehavior, 20, 141-183.2Diener, E., & Biswas-Diener, R. (2002). Will money increase subjective well-being? aliterature review and guide to needed research. Social Indicators Research, 57,119-169.3Heneman, R. L. (1992). Merit pay: linking pay increases to performance ratings. NewYork: Addison-Wesley.4Honeywell-Johnson, J. A., & Dickinson, A. M. (1999). Small group incentives: a reviewof the literature. Journal of Organizational Behavior Management, 19, 89-120.

5Jenkins, Jr, G. D., Mitra, A., Gupta, N., & Shaw, J. D. (1998). Are financial incentivesrelated to performance? a meta-analytic review of empirical research. Journal ofApplied Psychology, 83, 777-787.6Kasser, T. (2002). The high price of materialism. Massachusetts: MIT Press7Kasser, T., & Ahuvia, A. (2002). Materialistic values and well-being in businessstudents. European Journal of Social Psychology, 32, 137-146.8Kasser, T., & Kasser, V. G. (2001). The dreams of people high and low in materialism.Journal of Economic Psychology, 22, 693-719.9Luthans, F. (1973). Organizational behavior. New York: McGraw-Hill.10Luthans, F., & Kreitner, R. (1975). Organizational behavior modification. Glenview,IL: Scott, Foresman.11Milkovich, G. T., & Wigdor, A. K. (Eds). (1991). Pay for performance: evaluatingperformance appraisal and merit pay. Washington, DC: National Academy Press.12Nickerson, C., Schwarz, N., Diener, E., & Kahneman, D. (2001). The American dream:the dark side is in the wish, not the realization. Psychological Science, 14, 531-536.13Perry, J. L., Mesch, D., & Paarlberg, L. (2006). Motivating employees in a newgovernance era: the performance paradigm revisited. Public AdministrationReview, 66, 505-514. (Accessed on October 26, 2006at http://www.aspanet.org/scriptcontent/custom/staticcontent/t2pdownloads/PerryArticle.pdf)Money and employee motivation 9.14Sirgy, J. M. (1998). Matherialism and quality of life. Social Indicators Research, 43,227-260.15Skinner, B. F. (1969). Contingencies of reinforcement. New York: Appleton-Century-Crofts.16Srivastava, A., Locke, E. A. and Bartol, K. M. (2001). Money and subjective well-beingit’s not the money, it’s the motives. Journal of Personality and SocialPsychology, 80, 959-971.17Stajkovic, A. D., & Luthans, F. (1997). A meta-analysis of the effects of organizational

behavior modification on task performance, 1975-1995. Academy of ManagementJournal, 40, 1122-1149.18Stajkovic, A. D., & Luthans, F. (2003). Behavioral management and task performancein organizations: conceptual background, meta-analysis, and test of alternativemodels. Personnel Psychology, 56, 155-194.19Tolchinsky, P. D., & King, D. C. (1980). Do goals mediate the effects of incentives onperformance? Academy of Management Review, 5, 455-467.20Welbourne, T. M., & Gomez Mejia, L. R. (1995). Gainsharing: a critical review and afuture research agenda. Journal of Management, 21, 559-609.

Performance pay in the public sector: A review of theissues and evidenceUndertaken by:Graham PrenticeUniversity of WarwickSimon BurgessCMPOCarol PropperCMPONovember 2007The work described in this report was carried out under contract as part ofOME’s research programme. The views and judgements expressed in this reportare therefore those of the contractor and do not necessarily reflect those of OME.

Commissioned by: Office of ManpowerEconomics2

Executive summaryThe purpose of this report is to assess recent, robust empirical studies of public sectorincentive schemes and identify significant findings and common trends. As abackdrop we note some of the schemes in operation in the UK.Our assessment begins with the case for and against use of performance related pay,focusing on its use in the public sector and highlighting the fact that performance paycan have several effects, not all of which are beneficial.It then reviews the empirical literature for the following groups: civil servants(including police, armed forces and judiciary), healthcare workers and teachers. Inchoice of evidence, we have selected the most recent and the most robust evidence,

which often comes from exploitation of actual or natural experiments. We have alsotried to give some idea of the size of the evidence base for each set of workers, soassessment can be made of the ‘weight’ of the evidence. However, it should be notedthat overall the size of the evidence base is not extensive and much of it is not fromthe UK so that often we are reliant on a small number of papers.Our first conclusion is that there is evidence that public sector workers do respond tofinancial schemes, particularly in the field of education and possibly also in health.Sometimes these responses are small, but often the incentive in the scheme is smalltoo. Our second is that there is also evidence of gaming – manipulation of behaviourthat uses resources and does not increase productivity - in response to schemes. Ourfinal conclusion is that welfare calculations of the overall effect of performancerelated pay in the public sector are extremely rare, so that while it is possible to showthat in some cases public sector workers have responded to these schemes, the overallbenefits for society have not been assessed. In part, this is because such assessmentsare intrinsically hard to make; in other cases the schemes have not run for longenough to observe more than their impact on the employees who have been thesubject of the scheme.3EXECUTIVE SUMMARY.........................................................................................2INTRODUCTION........................................................................................................5SCHEMES CURRENTLY IN OPERATION ...........................................................5ISSUES ..........................................................................................................................6The case for performance related pay ..............................................................................................................6Productivity and efficiency ....................................................................................................................................6Productivity gains from recruitment – selection and sorting.............................................................................7The case against performance related pay: general issues ..........................................................................8Inter-temporal aspects.............................................................................................................................................8The case against performance related pay: particular issues for the public sector .............................9Multiple tasks, measurement and incentives........................................................................................................9Multiple principals .................................................................................................................................................10Teams/groups in production and rewards...........................................................................................................10Public sector motivation .......................................................................................................................................11RECENT EMPIRICAL EVIDENCE: .....................................................................12Why is it hard to get good (UK) evidence? ...................................................................................................12Private Sector Evidence ....................................................................................................................................13Civil Servants .......................................................................................................................................................14Summary .................................................................................................................................................................14The US JTPA programme ....................................................................................................................................15International evidence...........................................................................................................................................17UK evidence ...........................................................................................................................................................18Police – a special case...........................................................................................................................................20Armed Forces .........................................................................................................................................................21Judiciary ..................................................................................................................................................................22Healthcare Workers...........................................................................................................................................23Summary .................................................................................................................................................................234US evidence............................................................................................................................................................24UK evidence ...........................................................................................................................................................27

Teachers ................................................................................................................................................................ 29Summary .................................................................................................................................................................29US evidence............................................................................................................................................................30Evidence from Israel .............................................................................................................................................33UK evidence ...........................................................................................................................................................35CONCLUSION ..........................................................................................................375

IntroductionThe UK government has been reviewing the role of pay for performanceschemes (e.g. Makinson Report, 2000) and a number of public sector incentiveschemes are now in place at national and local level. But a recent report, (PWC 2006),concludes that little real progress has been made so far to link pay to performance formost public sector groups. Part of the reason for this is a policy of increasingcentralisation, which is making it difficult for local council managers in particular tocreate pay schemes in response to local market forces.This review first lists some schemes currently in operation in the UK. Themeat of the report begins with a brief summary of the case for and against the use offinancial incentive schemes, with particular focus on their use in the public sector.We then examine both UK and relevant international evidence on public sectorfinancial incentives. The evidence section is divided along common research lines toexamine civil servants, teachers and healthcare workers separately. Each main sectionstarts by summarising the key messages from the research and contains furthersubdivisions dealing with international and UK evidence. Some particular issues forthe armed forces, the police and the judiciary are also briefly considered. Our finalsection offers concluding remarks.

Schemes currently in operationPerformance pay remains part of the landscape of public sector pay, but is perhapslower down the political agenda than it was at the time of the Makinson Report. Afterthe report, there was a hiatus when some of the energy that went into reformingperformance pay went into a tougher stance on pay equality and pay ‘modernisation’(see IDS, 2007). More recently, however, HM Treasury identified performance pay asa key objective in their pay guidelines for 2006/7. Furthermore, performance pay is alive issue in discussions around teachers pay, and much of the controversy around thenew GP contract relates to the calibration of the points system for performance. Oneexample from IDS (2007) is from the pay settlement in the British Library, where“Treasury approval [for the deal] was forthcoming because the offer was seen to6balance a strong link between performance and pay, with froze n pay scale maximumsagainst a relatively high pay award and shortened pay scales” (p. 56). However,despite this, it remains true that for most in the public sector, the value of performancepay is modest. Some examples of this are as follows:In the Defence Aviation Repair Agency, “non-consolidated awards of between£50 and £150 were paid for performance” (IDS, 2007, p. 66).In the Department for Constitutional Affairs, “a flat-rate non-consolidatedbonus of £400 was paid to all staff awarded an ‘exceeded’ performance rating”(IDS, 2007, p. 68)In the Foreign and Commonwealth Office, non-consolidated rewards forexceptional performance varied by salary grade between £900 and £1850.

(IDS, 2007, p. 76)In the Home Office, top ranked performers were awarded 2% of salary. (IDS,2007, p. 80)In local government, performance pay also remains rare (IDS, 2007, p. 100,107).Pay for performance structures in education and health are discussed in therelevant sections below.

IssuesThe case for performance related payProductivity and efficiencyThe arguments for linking pay to performance are the following. Effective financialincentives provide an opportunity to improve the productivity of public sectorworkers. With compensation linked to performance, employees should expend moreeffort, lifting the quantity and/or the quality of their output. Thus by promoting betterperformance internally, governments can use incentives as a means of deliveringsuperior public services. The introduction of performance related pay can alsomotivate employees to pursue professional development opportunities that previously7offered little in the way of additional benefits for the individual. Productivity istherefore likely to improve both in the short run, because employees are workingharder, and also in the longer run, as staff professional development generates furthergains in productivity (see Lavy 2007).In the public sector, financial incentives also project a clear message aboutwhich outcomes are valued by society, and by how much. Employees can thenprioritize tasks correctly and allocate more time and effort to higher valued tasks. Thisis the so-called “efficiency” case for performance related pay, which demands thatincentive scheme designers have a detailed understanding of the priorities of the usersbeing served.Productivity gains from recruitment – selection and sortingAlongside improvements in the productivity and efficiency of current public sectoremployees is the impact of financial incentives on the potential pool of new applicantswaiting to join. Certainly, increasing the level of compensation associated with suchwork will increase the quantity and quality of new applicants. Offering the chance ofhigher pay for strong performers might tempt such cand idates away from what areotherwise more attractive opportunities elsewhere. This “selection” effect helps tobring more able workers into the workforce, with obvious positive consequences foroverall productivity. In fact, Lazear argues that as a source of productivity gains, theselection effects from performance-linked-pay are similar to the gains available frommotivation effects, if not greater (see Lazear 2000).Financial incentives (and penalties) could also have desirable consequencesfor the retention of high achievers and the turnover of low achievers over time. Besleyand Machin (2006) found that the existence of a performance premium associatedwith being a good head teacher resulted in poor head teachers leaving their jobs ratherthan accepting lower salaries. Lavy (2007) notes, however, that equally poor outsideopportunities might curb this kind of effect for lower level workers and moregenerally, the presence of (or lack of) alternative opportunities will have a significantimpact on the sorting of workers between establishments.8The case against performance related pay: general issues

Inter-temporal aspectsMost incentive schemes remain in place for extend periods during which theperformance of employees is measured and bonuses are awarded multiple times.Employers can, therefore, observe the ability of an employee from previousperformance measurements and make an assessment of the difficulty of the work.Employers wishing to reduce the costs of their scheme can then realign thresholds andraise targets to restrict the distribution of financial bonuses at all performance levels.Anticipating such changes, employees are likely to reduce their effort inearlier periods to mislead the employer into believing that the task is more difficultthan it is in reality. This phenomenon is known as the “ratchet effect”, which occurswhen employers give an undertaking not to modify the incentive system after it isintroduced. Like free rider effects in large teams, ratchet effects have the potential toreduce the value of financial incentives schemes to employers (see Burgess and Ratto2003).In contrast, career concerns can work in the opposite direction to ratcheteffects. Whether ratchet effects can be overcome depends upon the age of employees,their prospects and the level of uncertainty regarding their ability, amongst otherfactors. Supposing an employer is unaware of the true ability of an employee.Knowing this, the employee can work harder initially, to form the impression thatthey are a high ability worker. An employer will then observe the employee’s higheffort level and form the perception that they are a high ability worker. The pay offfor the employee emerges over the long run, in terms of improved job opportunitiesand higher future earnings. Indeed, the existence of career concerns suggests thatexplicit financial incentives may not be necessary to motivate younger, untestedemployees. Furthermore, greater uncertainty regarding employee ability and trueperformance, as is a feature of public sector roles, may actually strengthen careerconcerns effects (see Dixit 2002). Attempts to measure performance in order to link itto pay may, therefore, reduce these effects.9The case against performance related pay: particular issues for thepublic sectorWhile some of the possible issues relating to financial incentives are generic, thespecial nature of the public sector creates particular challenges.Multiple tasks, measurement and incentivesIn many cases public services are multifaceted and, as a result, the objectives of anorganisation are difficult to define and so to measure. So for example, the objectivesof a school might be to provide a “good education” but this is harder to define than,say, the production of a cars or the adequate collection of garbage. Even breakingdown the overall objectives into sub-components can be difficult; there is forexample, much debate about what constitutes a good measure of hospital quality. Thismeans that it can be hard to find good measures of performance and the measuresadopted may be very noisy – they impart relatively little information about the effortof an employee or the organisation (Propper and Wilson 2003). As a consequence,linking rewards to the meeting of performance targets does not give effectiveincentives, may lead to some elements of strategic behaviour and imposesunnecessary risk on employees.One of the reasons that these services are complex is because they involveseveral dimensions, some of which are relatively easy to measure, others of which aremuch harder to measure. Examples of the former are school students’ performance on

standardised tests and the re-admission of patients after operations. Examples of thelatter are becoming culturally aware or improving the long term health of patients.This difference in measurability may mean that incentives can only be linked to theeasy to measure outcomes. This may lead to an excessive focus on these outcomes atthe expense of other tasks. So hospitals which receive financial incentives linked tocost reduction may have incentives to compromise on quality, while teachersmonitored on student pass rates may choose to reduce the effort they put into less ablestudents.Exerting more effort on one task, therefore, reduces the amount of time oreffort available for other tasks which are substitutes. In this way, higher incentives in10one task will drive an employee’s effort away from alternative tasks. To avoidmisallocation of effort by an employee, the employer will have to weaken theincentives on the more accurately measured tasks.Multiple principalsPrecisely because the services are public in nature, several groups in society will beinterested in their provision, cost and efficiency. The tax payer is always one party:tax payers’ desires to keep down costs may conflict with those who are providing theservice who are interested in high quality. Within an organisation, public serviceproviders may serve several masters: for example, doctors get professionalrecognition from their research activities, but the managers (and possibly the patients)of hospitals might prefer they focused their time on other, less professionallyrecognised tasks, such as reducing waiting times for common procedures. Theeconomics literature stresses that in these cases - referred to as multiple principals -financial incentives should be weaker than in the case where there is only a singleprincipal.Teams/groups in production and rewardsPerformance incentives designed to reward performance at group level have become avery popular way of rewarding good performance in the public sector (e.g theMakinson schemes in the UK). Team level financial incentives are attractive partlybecause they help to address concerns that individual level incentives might lowermorale and, ultimately, reduce productivity in the workplace. Another reason for theirprevalence is that group level performance measures are often the only ones available;measuring public sector output at an individual level can be difficult. Indeed, they arepopular despite the fact that group level schemes ought to suffer from “free-rider”effects that could reduce the impact of financial incentives on performance. The freeriderproblem occurs when the work of each team member complements the work ofothers to the extent that each person’s contribution is indistinguishable. Large teamsizes are more likely to suffer from free-rider effects, as are teams that are involved inthe production of output that is hard to measure or quantify (as is commonly the case11in the public sector). More expensive monitoring techniques may be necessary, andthis can add to the costs of implementing a financial incentive programme.On the other hand, team incentives can help to promote peer monitoring insmaller teams, particularly when team members are mutually dependent on each otherto achieve performance thresholds. Whether or not the threat posed by peermonitoring is credible will depend on how long the team is together. Teamrelationships are built up over time and repeated interactions amplify the costs ofweak workers to teams. Thus under team-based financial incentives, co-operation andinterdependence could provide a motivation to exert peer sanctions.

Public sector motivationMuch as career concerns provide an internal motivation to exert extra effort even inthe absence of financial incentives, public sector (intrinsic) motivation can have asimilar impact, this time independent of an employee’s age and the level ofuncertainty over job performance. Pub lic sector workers may care about the outcomesor mission of the public organisation and gain satisfaction, indeed, intrinsiccompensation, from knowing that their work is contributing to it. Having internalisedthe goals of their organisation, police officers may care about the safety of citizens,nurses and doctors about the health of their patients and teachers about the attainmentof their pupils, to the extent that they obtain welfare from seeing their user’s needsbeing met. It is possible that external financial incentives could overwhelm publicservice motivation, since it suggests to the employee that their employer recognisesno association between output and effort other that a pure, market relationship.There are two recent developments to this argument. Francois (2000) suggeststhat intrinsically motivated employees should actually work best when incentives aresmall or even absent and employers commit not to divert any surpluses or publicsector “profits” away from the organisation’s mission. Besley and Ghatak (2005)develop this further, arguing that that if public sector organisations post missionsduring their recruitment process, the natural sorting of applicants will do the job offinancial incentives. On the other hand, financial incentives may help to focus efforton other organisation goals which could have been neglected if employers relied onpublic service motivation alone. This brings us back to the efficiency argument in12which financial incentives help well meaning, intrinsically motivated employees toprioritise tasks in the “right” way.

Recent empirical evidence:Why is it hard to get good (UK) evidence?While there is a large literature on financial incentives in the private sector,particularly at CEO level, actual empirical evidence forms only a relatively small partof this. In the public sector, much less research into financial incentives has beenundertaken, and empirical evidence is particularly scarce. In order to present a morecomprehensive, informed view, it is necessary to consider the UK literature (where itdoes exist) alongside evidence from financial incentive programmes operating inother countries. Even so, large gaps in the research remain, particularly with respect tothe armed forces, the prison service, nursing and midwifery, and at senior levels. Thetable below highlights some of the gaps in the empirical literature and specifically, thelack of substantial UK evidence.The table details the number of empirical studies on public sector financial incentiveprogrammes found by profession and region dated 1999 or later:TeachersCivilServantsDoctors Police Judiciary MilitaryUK 2 2 3US 7 5 7 4 1Othercountries8 1 2This lack of evidence partly arises because of the inability or unwillingness of

governments to carry out experiments within the public sector: changes are eitherintroduced for all service deliverers or all eligible recipients, or they are notintroduced at all. This means that it is difficult to net out the effect of a scheme fromother changes that may have taken place at the same time as the scheme. So much of13the early evidence is not very robust. Consequently, a meta-analysis in which it isconcluded that the “bulk” of the evidence shows this or that cannot be provided as –despite the strong positions taken by the proponents or opposers of attempts toincentivise the public sector - there is no bulk (Burgess and Ratto 2003).In selecting evidence, therefore, we have focused on the more robust analysesthat is taken from either actual experiments or from so-called ‘natural experiments’:situations in which those subjected to performance related pay can be compared to anappropriate, unaffected, control group. This method has the added advantage thatcharacteristics such as “ability”, which may have a bearing on the results but which,due to their nature, are unobservable, can be factored out using the data. Thedisadvantage of many such experiments is that participants are often not chosen atrandom and so are not representative of their sector. This remains a common problemthroughout much of the empirical evidence on public sector financial incentives.Private Sector EvidencePrendergast (1999) remains the most recent, comprehensive review of robust evidenceon incentives in both private and public sectors. It concludes that:There is strong evidence that workers do respond to incentives;Job contracts, in which the type of incentive being used by the employer isspecified, frequently utilize a different type of incentive structure to the onepredicted as optimal by economic theory. The implication of this is that someof the economic theory of incentives remains unconfirmed by empiricalevidence;Evidence of incentive programmes that reward relative performanceimprovements is also inconclusive;The requirements of empirical literature, that output and quality is easy toobserve and measure, mean that very little evidence exists for the majority ofjobs in which performance is measured using subjective performanceevaluations. This represents a substantial gap in the literature.14A robust private sector case study comes from Lazear (2000) who analyses theresponse of 3000 employees of Safelite Glass Corporation (a large US autowindscreen installation firm) to a new contract that links workers’ pay to the numberof windscreens they install. As the new contract was phased in gradually between1994 and 1995, Lazear is able to use individual level data covering a 19 month periodto control for outside influences on worker productivity. He estimates the switch fromhourly wages to output-linked (piece-rate) pay resulted in a 44% gain in output perworker on average. He found that incentive effects were responsible forapproximately half the 44% gain in output, while the other half of the gain wasattributed to selection effects. Further analysis suggests the large selection effect cameas a result of a reduction in employee turnover among high performing employeesalready employed by Safelite, and from the recruitment of more able workersattracted by the potential to earn higher wages. The scheme resulted in a 10% wageincrease for workers. The firm addressed quality issues by adopting a policy wherebythe worker responsible for a poor job (as indicated by a windscreen that broke) had toreplace the windscreen without receiving wages for it. As a result of this policy,

Lazear actually identified a slight improvement in quality after the incentive schemewas introduced.Other studies by Paarsch and Shearer (2000), Shearer (2003) and Ba ndiera,Barankay and Rasul (2005) all suggest private sector firms in similar industries (treeplanting and fruit picking) witness gains of a similar order of magnitude to thosefound by Lazear (2000).Civil ServantsSummaryAll the studies examining incentive schemes for civil servants evaluateprogrammes that reward work at group or team level.There is strong evidence that civil servants do respond to financial incentives,even when benefits are not directly salary enhancing.15One UK study suggests that team incentives might be more effective insmaller groups in which performance can be easily monitored by colleagues orpeers.In several US studies, strategic behaviour was identified amongst civilservants in incentive programmes. One paper suggests that this kind ofbehaviour is more likely when performance is measured inaccurately or lessaccurately relative to other performance measures.UK evidence comes from programmes which were only in place for a limitedtime period of one year or less. The short duration of these schemes makes itunlikely that the novelty of incentives will have worn off. The long termresponse to incentives may not, therefore, be the same as short term impactssuggest.Most studies have not considered whetherfinancial incentives represent goodvalue for money overall, although one US paper provides evidence thatstrategic behaviour by civil servants was placing identifiable costs on society.The full welfare implications of financial incentives for civil servants,therefore, remain unclear.One further complication is the need to identify appropriate performanceincentives which let civil servants prioritise tasks in the “right” way. Thisproblem is particularly acute for the police and armed forces.The judiciary is one group for whom financial incentives are traditionallyconsidered inappropriate.The US JTPA programmeA rich source of evidence comes from the US Job Training Partnership Act (JTPA), afederal employment and training programme which promotes desirable employeeconduct using agency level financial incentives rather than outright regulation. In thisscheme, the US Department of Labor awards bonuses directly to the training centresand these bonuses cannot be used to supplement employee salaries. Thus whilesocially motivated employees may obtain benefits from better facilities and observinghigher staff numbers, individual performance is not rewarded directly. The actual size16of the reward depends on an agency’s performance according to a range of measures,which have varied over the life of the programme and from State to State.Performance was originally measured in terms of cost per participant as well asparticipant wage rates, employment status, and earnings up to three months after theygraduated. Bonuses have augmented the operating budget of agencies by 7% onaverage (Courty and Marschke 2004).

Common findings from empirical studies of the JTPA programme aresummarised in Barnow and Smith (2004). They first find mixed evidence overall of‘cream-skimming’, the practice of selecting applicants most likely to help agenciesmeet their performance targets. Heckman and Smith (2004) look at 16 agencies and,by decomposing the selection process, find that lack of awareness of schemeeligibility rather than cream-skimming by agency staff is responsible for driving mostof the observable differences between eligible candidates and selected participants.Examining one agency in detail, Heckman, Smith and Taber (1996) find evidence thatagents are deliberately choosing less employable candidates despite the presence ofperformance related financial incentives. Barnow and Smith (2004) suggest thatregression-based methods to adjust the performance measures used in the schemes forcase mix have been somewhat successful at reducing the incentives to cream-skimattractive applicants.Second, in contrast to the inconclusive results on cream-skimming, there issubstantial evidence of strategic behaviour, in which agents improve measures ofperformance without increasing their actual performance. Courty and Marschke(1996, 1997 and 2004) look at the choice of termination date made by agencyworkers, on which day the key performance measurements are taken. Under thescheme, agents can choose any date up to 90 days from the end of training toterminate a participant and record their employment and salary details. Since bonusesare awarded on the basis of annual data, agents can engage in strategic reporting tospread good and poor performance over multiple years. This behaviour rewardsagencies because bonus payments are target based, with no additional benefitsassociated with exceeding targets.17In their study based on 16 agencies, Courty and Marschke (2004) first showthat agencies delay for longer the termination date of programme participants if theyremain unemployed throughout the 90 day period. Secondly, they identify an annualpattern in their results, in which unemployed participants are terminated earlier in thelast three months than those finishing within the first 9 months of the programmeyear. Then thirdly, looking at their findings for the last three months alone, theynotice that both very good years and bad years are associated with earlier programmeterminations for unemployed participants. This is what we would expect to find ifagents were engaged in strategic behaviour, as agencies which are either unlikely toreach their target performance, or have generously exceeded it, have nothing to loseby terminating unemployed participants within the current programme year.Whether this strategic behaviour imposes welfare costs on service users andother stakeholders is also examined. Courty and Marschke (2004) compare theearnings of programme graduates from agencies with high and low levels of strategicbehaviour and find high levels are associated with lower graduate earnings. Further,they observe that courses in June - the final month of the programme year - are morelikely to be truncated than in other months, and that those participants taking Junecourses generally show a smaller wage impact from the programme. This is allevidence that strategic behaviour in this environment, far from simply being anaccounting issue, is also imposing real costs on society as well.International evidenceSome of the strongest evidence that financial incentives can influence the output ofcivil servants comes from Khan, Silva and Ziliak (2001), who evaluate theintroduction of a performance pay scheme within the Brazilian Tax CollectionAuthority. The scheme, initiated in 1989, rewarded both individual and group effort

using objective measures of relative efficiency including the number of inspectionsmade and the amount of fines collected from tax evaders. This was supplemented bysubjective performance appraisals for individuals. The scheme paid bonuses equal to70% of additional fines collected, and as a consequence, substantial bonuses reachingtwice the mean annual salary became common. That this bonus scheme was so highpowered makes this case study relatively unique.18After netting out region specific characteristics, Khan et al. (2001) evaluatethe impact of the scheme from data covering three years before and three years afterperformance pay was implemented. Overall, they estimate that the bonus scheme ledto a 75% increase in the rate of growth of fines per inspection. At the same time, theyalso identify substantial regional variation, with responses ranging from 19% to145%. The authors suggest that diverse management techniques resulted in someregions targeting wealthier sources (such as corporations), more aggressively. Thisprompts Khan et al. to issue a warning that high powered incentive schemes such asthis risk encouraging extortion in their workforce and may have significant socialcosts as well as clear financial benefits. Unfortunately, limited data prevents themfrom examining these social costs further.UK evidenceHere, the evidence is limited to only a handful of studies. Burgess et al (2004)examine the impact of an incentive programme piloted in Jobcentre Plus, a UKgovernment employment and benefits agency. Like the US JTPA programme, the UKJob centre pilot was team-based, with performance assessed at team level. In all, 90teams took part, consisting of between five and 39 offices ranged in size from 264 to1535 people. Unlike the US scheme however, bonuses were paid to workers directly,with each team member receiving a 1% increase in salary for each performance targetmet. Meeting all five quality and quantity targets was associated with an additional2.5% increase in salary, so workers in a high performing team could benefit from asalary rise of 7.5% in total. Crucially, quantity was measured in terms of job entrypoints (JEP) achieved, with for example, the successful placement of a jobless loneparent attracting twelve points, compared to one point for an already-employedservice user. Further points were awarded if the service user was still in employmentafter four weeks.Burgess et al. (2004) focus on three aspects of the Jobcentre Plus incentivesscheme: first, the extent to which financial incentives influence the behaviour ofpublic sector workers; second, the consequences of designing an incentive schemethat rewards team rather than individual effort, and finally, whether a wide variation19in the accuracy of performance measures can influence the behaviour and effort ofworkers in the public sector.Looking at job placement performance, Burgess et al. find that while theoverall response to incentives was close to zero, team size made a significantdifference, with ‘very small’ and ‘small’ teams showing 10% increases in outputcompared to evidence of reduced output for ‘large’ teams. They interpret this assuggesting that peer performance monitoring, more easily done in smaller teams, iseffective at reducing the problems associated with team-based incentives.Furthermore, for smaller teams, the small additional cost of the incentive schemecompares favourably with a general pay rise or an increase in staff numbers. Splittingperformance into quantity and less easily measured quality components, the schemeappeared to result in improvements in the quantity but not the quality of services

offered. This suggests that measurement accuracy does indeed have an importantimpact on worker’s strategic behaviour in this context.In another UK study, Burgess et al. (2005) evaluate two different team-levelfinancial incentive schemes in operation at HM Customs and Excise during 2002 andcompare their performance to a randomly selected group that was not involved ineither bonus scheme. Both teams were assessed on their performance in identicaltasks, included auditing work and tax revenue collection. One team had bonuspayments that paid a fixed percentage of officers’ salary (ranging from £550 to£1520) while the other team had flat rate bonuses which were worth £740 for eachindividual. Teams were required to meet all non- incentive targets in addition tofurther targets on both key performance indicators in order to receive their respectivebonuses.The main aim was to investigate whether and how individuals changed theirbehaviour in response to team-based incentive schemes. Overall, Burgess et al. (2005)find that officers of both teams worked harder over the period than the randomlyselected control group. However, only team two (the fixed bonus team) succeeded inhitting all its targets. Seeking to explain this difference, Burgess et al. find evidencethat managers in team two allocated more incentive tasks to efficient officers (whowere among the top 25% most productive workers before the scheme was initiated)20than managers in team one. Whether the flat bonus structure led to the strategicbehaviour by managers in team two is not addressed. Hence, while all officers in bothteams responded positively to the incentive programme, the team facing a uniformfixed bonus also worked smarter as a team, strategically deploying efficient officerswhere their effort was best rewarded.Police – a special casePolice forces represent one government bureaucracy in which it is particularlydifficult to identify appropriate performance incentive measures. Providingperformance incentives for some of the measurable outcomes that do exist could haveunintended consequences which might reprioritise police tasks against the wishes oflocal communities (see FitzGerald et al. (2002) cited in Vollaard, 2003). Indeed,Prendergast (2001) offers evidence from the Los Angeles Police Department thatincreased performance measurement and public oversight on their own underminedofficer’s willingness to make arrests, which in turn led to a significant increase incrime rates.Baicker and Jacobson (2007) examines the police response to financialincentives introduced by the US Department of Justice. Initiated in 1984 as part of theComprehensive Crime Control Act, the programme provided police departments theopportunity to share a proportion of the proceeds of drug-related asset seizures.Actual proportions were determined by State governments, resulting in a widevariation in the fraction returned to police departments. Netting out differences incounty characteristics, Baicker and Jacobsen estimated a 10% increase in the fractionreturned to police departments is associated with a $0.19 per capita increase in thevalue of seizures police made. States which returned almost all drug related assetseizures added a further $0.09 per capita to this figure. Police departments retaining ahigher fraction of their seizures also worked more strategically, placing greateremphasis on possession offences than drug sales, which are less likely to include asignificant cash component on arrest.While this particular scheme has been relatively successful, it remains highlyunlikely that this programme design can be replicated across the full range of police

duties. A further problem arises from the difficulty of finding accurate quality21indicators in order to prevent officers from meeting targets without raisingproductivity. Changing regional and national priorities and the need for detailed localknowledge also implies that targets will quickly become inappropriate. Subjective,qualitative performance appraisals may therefore be more effective in this context.Armed ForcesThe armed forces encounter many of the same measurement problems faced by policeforces. Partly as a response to this, military forces have developed a system in whichpromotion confers greater status and privileges, (in addition to the higher salaries andallowances typically found in civil employment). These non- financial rewards providesubstantial performance incentives on their own, for as long as individuals remainemployed in the armed forces. Even so, the main purpose of the promotion system inplace continues to be aimed at identifying personnel able to perform well in rolesassociated with higher levels of responsibility, as opposed to one explicitly designedto reward performance. Hence, the link between performance and promotion is notnecessarily a straightforward one –promotion depends upon being given sufficientopportunities to demonstrate ability including educational background and commandresponsibilities, in addition to ability itself.Under the current system, high performing individuals have the opportunity torise up the ranks quicker than average performers, while consistently poor performersare more likely to be discharged under an “up or out” structure. Increasing the extentto which pay is linked to rank rather than years of service is one way of making thestandard armed forces reward programme more effective at providing performanceincentives. Since the chances of promotion falls as military personnel approachretirement age, the incentive effect associated with promotion is also likely to fallunless the incremental increases in pay associated with the highest ranks are largeenough to counter the increased risk. Thus, while no evidence of a performancerelated pay schemes exists for armed forces, elements of a merit pay programme arepresent or can be made more explicit within the current compensation framework.Asch (1993) and Kosiak (2005) discuss current issues in US military pay.22JudiciaryFinancial compensation policy for the judiciary is designed to remove all externalpressures in order to allow judges to act independently. This independence is a vitalcomponent of any democratic political system, and any compensation policy that addsexternal pressures could, therefore, be inappropriate. Addressing the US system, Choiand Gulati (2004) propose the creation of a set of objective performance measureswhich can then be used to rank judges in order to identify candidates suitable forpromotion – up to Supreme Court level. This induces incentives for judges to improvetheir performance ranking. Their justification for such a tournament is that the presentsystem is excessively politicised and improved ways of measuring and consideringjudicial quality need to be considered. Examples of possible performance criteriainclude the publication rate of opinions, their citation rates by other courts, citationrates by the Supreme Court, dissent rates, case productivity etc. It is possible thatequivalent objective performance measures do exist for the UK, but identifyingappropriate measures and their weightings could be difficult and any strategicbehaviour by judges other than productivity improvements in judicial quality couldhave serious consequences.Ramseyer and Rasmusen (1997) look at all Japanese judges who started work

between 1961 and 1965 in order to identify key determinants of career success and theextent of political influence in the Japanese judicial system. The Secretariat grades theproductivity and quality of all judges and uses this performance information toallocate jobs of greater or lesser prestige and pay. The system enables the Secretariatto reward ability and provide performance incentives for judges who begin theircareers unproven and at a comparatively young age. While Ramseyer and Rasmusendo not examine whether judges respond to financial incentives, they do find evidencethat judges responded to some incentives which were politically biased. Theincumbent political party was, therefore, able to use the system in place to influencethe judiciary without overtly intervening in the decision-making process.23Healthcare WorkersSummaryImproving quality is the primary focus for many, if not all, performance payschemes implemented in the healthcare sector. This stands in contrast to civilservants’ financial incentive schemes, which attempt to improve the qualityand quantity of output.The evidence that doctors respond to financial incentives is inconclusive.However, much of the US evidence comes from incentive schemes whichwere only in place for a limited period and in nearly all cases, the financialrewards involved were small.There is limited evidence from US studies that a clinician’s response toincentives depends upon the absolute and relative size of the financialincentives as well as the relative effort required to obtain them.UK performance measures suggest that the quality of healthcare delivery fromGPs is generally very high. GPs have therefore been very successful inobtaining financial bonuses under the current performance pay scheme.However, this is against a backdrop of previously high levels of performancefrom GPs – so the extent to which this scheme has been successful is unclear.Both of the UK studies looking at GPs’ response to incentives identify a smallbut significant proportion of doctors who may be deliberately misreportingperformance measures to increase their financial rewards.The analysis from one of these UK studies also suggests that GPs aremotivated by concerns for patient health as well as financial incentives.Unfortunately, the short duration of many incentive programmes meansevidence on cost effectiveness is scarce. Care must, therefore, be taken whenequating higher performance measures with actual improvements in the healthof society.24US evidenceA wide ranging US study by Lindenauer et al. (2007) evaluated a pilot programme offinancial rewards based on rankings. This started in 2003 for two years, operatingalongside a public reporting programme. Both programmes were developed to pushthrough improvements in healthcare quality. The authors compare 207 hospitalswhich participated in both schemes to 406 others which have matching characteristicsbut were only taking part in the public reporting programme. In this programme,hospitals are ranked according to their absolute performance in 33 indicators ofquality (as a percentage of patients received for that condition). Hospitals in the topdecile overall received a two percent budgetary supplement while those in the seconddecile received one percent. Hospitals with fewer than 30 cases of a particular

condition annually were excluded from related quality indicators. The scheme waspartially funded by financial penalties for poor performing hospitals.After taking into account differences in initial performance levels and volume,Lindenauer et al. (2007) attribute the financial incentive scheme with achievingimprovements of 2.6% and 4.1% above that of non- incentive hospitals over the whole2-year period. Furthermore, they found some evidence that the largest improvementswere being made by what were, initially, the poorest performing hospitals. Theseresults suggest first, that healthcare staff do respond to indirect, quite modest,financial incentives. Second, a scheme of financial incentives and penalties can elicitimprovements in reported quality from hospitals which are already publicly reportingtheir performance indicators. Finally, they note that this scheme saw measuredimprovements in quality indicators for both low and high performing hospitals, eventhough bonuses were only paid to the top 20% of hospitals. However, the extent towhich these results can be generalised for all US hospitals is limited by the fact that 1)hospitals in the non-incentive group chose to opt out from the study (possiblyintroducing some bias to the results) and that 2) the characteristics of the incentivescheme-hospitals are not typical of US hospitals as a whole.Much more common than rank-based incentive schemes in the healthcaresector is performance targeting. Under targeting, performance is rewarded accordingto a hospital’s or practice’s ability to meet pre-defined thresholds. Rosenthal and25Frank (2006) conducted a survey of the US empirical literature on performance pay inlate 2003. They identified one rank-based incentive scheme and six performancetargeting programmes. Their general conclusion is that, as of 2003, empiricalevidence that financial incentives can improve the quality of healthcare is weak. Inparticular they are critical of the research design of a number of studies which findsignificant positive effects of financial incentives. However, Rosenthal and Frank alsoacknowledge that bonuses may have been too small (an incentive scheme evaluatedby Hilman et al. (1998) awarded top level bonuses of only $1260 per site) and that thepresence of multiple contracts placed too many competing demands on physician’swork. Conditions are very different in the UK, where hospitals have fewer contracts.A later survey of the empirical literature on performance pay in health carewas carried out in November 2005 by Petersen et al. (2006). Comparing the results of17 different US studies, they find tentative evidence that pay for performanceschemes have had positive effects. As with Rosenthal and Frank (2006), however,they identify limitations to the literature which prevents them from drawing anythingother than preliminary conclusions. These conclusions are:Financial incentives often induce better book keeping and documentationmethods than actual increases in medical provision (which suggests poorlydesigned performance targets).Incentives are more effective when the potential reward is larger and thepayment frequency is higher.Threshold based systems are failing to reward organisations which improvethe most because they often have the lowest baseline performance.Rosenthal et al’s (2005) evaluation of a group-level physician threshold schemeoperating in California provides an example of this type of reward distributionproblem happening in practice. The QIP scheme, initiated in 2002, offers financialbonuses for each target of $0.23 per member per month if the physician groupexceeded the performance achieved by the 75th percentile in previous year. Theaverage physician group had about 10,000 continuously enrolled members, but even

so, a perfect performance in all 10 quality targets would only yield a bonus equal to0.8% of an average group’s annual revenue. The 163 practices in California were26compared over time to 42 similar practices in Oregon and Washington State whichdid not take part in the scheme. Looking at three quality targets for which they havepre and post programme data, Rosenthal et al. find a statistically significantimprovement (a 3.6% difference) for just one of the three performance measures.Looking at which groups yield the highest bonuses, they find that 75% of bonuseshave been awarded to physician groups which were initially performing above targetlevels, even though these groups had shown the smallest improvement.Later on in January 2003, these same Californian practices were submitted to anew, parallel incentive scheme with approximately 10 times larger financialincentives (the HIA programme). This new HIA scheme cost six US health plans$122.7m in 2004 and $139.5m in 2005. Mullen et al. (2007) evaluate the impact ofboth schemes together using data before and after the schemes were introduced (July2001- July 2005), comparing their results to Washington and Oregon practices whichdo not participate in either programme. They also examine quality indicators ignoredby both schemes.Their results show a positive response to some of the rewarded performancemeasures which do not demand significant additional doctor time/effort. Examples ofthese include cervical cancer screening and childhood immunisation rates - indicatorswhich rely on identifying high risk groups and scheduling appointmentsappropriately. Indeed, for these indicators clinicians respond more positively underlarger incentives. In contrast, the prescription of preferred antibiotics, which is onlyrewarded by the weaker QIP scheme, shows significant declines of between 2.5% and10% when the HIA program is introduced. Among unrewarded performancemeasures, the response was more mixed, with two showing above trendimprovements while a third indicated falling performance. Despite some variation inquality of care indicators in response to incentives, Mullen et al. (2007) do not findany evidence to suggest that overall patient health has suffered from the introductionof the QIP and HIA incentive programmes, but they do question whethe r the schemewas cost effective, given mixed and unpredictable results.27UK evidence2004 witnessed the introduction of a UK-wide financial incentive programme forgeneral practitioners which was both broad-based and high powered. The governmentcommitted £3.2bn to the programme in its first year, potentially augmenting generalpractice (GP) budgets by 25% on top of their core funding allocations. Theprogramme rewards the performance of family practitioners based on 146 qualityindicators, measured at practice level. These indicators cover various aspects ofclinical care, practice organisation and patient experiences. Rewards are weighted fordifficulty, and are awarded on a sliding scale of achievement up to a maximum target,above which the bonus stays fixed. In this way, the UK pay for performanceprogramme does attempt to address some of the reward distribution issues identifiedby Rosenthal et al. (2005).Even before the introduction of the scheme, there is evidence that physicianswere responding to quality initiatives. Campbell et al. (2005) follow three quality-ofcaremeasures (coronary heart disease, asthma, and type-2 diabetes) for 42 generalpractices between 1998 and 2003. Using patient level data aggregated at practicelevel, they observe significant improvements for 22 of the 50 quality indicators

relating to these conditions. Possible explanations for this improvement includeadvances in data recording techniques, as well as an expansion of public reportinginitiatives focusing on quality. Campbell et al. predict high levels of performancemeasured under the new scheme, and that has since been confirmed in studies byDoran et al. (2006) and Gravelle, Sutton and Ma (2007).Sutton and McLean (2006) examine a selection of performance measures for60 clinical practices in north Scotland. These practices scored on average 612 pointsout of a maximum 650 points in 2004. Analysing which factors may be influencingquality measures they find that size and team composition are the most significantdeterminants of performance. According to the study, practices with more that fourfull-time GPs (clinicians) and practices with younger staff obtained higher qualityscores. Thus larger practices performed better, a result which is seemingly unrelatedto the proportion of nursing staff at a practice. Confounding concerns (and evidencefrom Gravelle, Sutton and Ma, 2007) that practices in more deprived areas might28suffer under the new system, those serving more materially deprived populationsactually achieved higher scores. Former fund holding practices and training centresalso scored better.Two studies actually evaluate the effects of the performance programme ondoctor behaviour. Doran et al. (2006) focus on 76 clinical quality indicators for 8105GP practices (95% of practices) in the UK. They observe that the average practiceearned an additional £76,200 from the scheme for the 2004-2005 reporting year, with230 practices achieving the maximum bonus. In contrast to Sutton and McLean(2006), analysis by Doran et al. suggests that smaller practices perform better,although larger practices perform more consistently. They do, however, both agreethat practices with younger doctors achieve better scores. In addition, Doran et al.consider rates of exception reporting for special cases and conditions. Doctors canreport an exception for special cases, when the appliance of performance targetsmight lead to inappropriate treatment. They observed that levels of exceptionreporting were low across the UK (with a median rate of 6%), but this included 91practices (1.1%) which reported exceptions for more than 15% of their patients.Furthermore, exception reporting was also an important determinant of performance,with 1% higher rates associated with an increase in measured performance of 0.31%.A possible explanation for this result is that these practices have been using exceptionreporting to obtain higher bonuses without actually improving outcomes, referred toas ‘gaming’ behaviour in the economics literature.Gravelle, Sutton and Ma (2007) investigate the effects of exception reportingfurther in their study, which examines a sample of 916 Scottish GP practices underthe same pay for performance programme. The authors note that the incentiveprogramme measures performance as a ratio of patient treatment divided by conditionprevalence. So a practice performing below the maximum threshold in any onequality indicator can increase their reward bonus by:Increasing the rate of exception reporting,Reducing reported levels of condition prevalence in the local population orIncreasing the amount of patients appropriately treated.29Perversely, under the NHS incentive scheme, over-performing practices also have anincentive to report higher prevalence rates. Gravelle, Sutton and Ma follow 65 clinicalquality indicators for each practice over the 2004/2005 and 2005/2006 reporting yearsin an attempt to identify any changes in doctor response to incentives.

Their results show that some practice characteristics, such as the proportion offemale clinicians or younger clinicians were positively associated with higher qualityscores and higher prevalence rates, a somewhat unexpected result. The authors alsocalculate that GPs chose to treat 12.4% more patients than were necessary tomaximise their financial rewards, suggesting clinicians are motivated by more thanjust financial incentives. However, in other analysis, GP practices that exceededmaximum performance thresholds during the first year are significantly more likely toincrease reported prevalence rates the following year. Practices which performedbelow the maximum thresholds in 2004/2005 are also found to have increased theirexception reporting for 2005/2006. These results suggest that, while most doctors arebehaving quite altruistically under the new NHS incentive scheme, a small proportionof doctors have been taking advantage of the system, reporting higher performancethan was actually delivered in order to enhance their financial reward.TeachersSummaryIncentive programmes in the education sector can typically be divided into threecategories:1. Merit pay schemes reward pupil performance on a subjective basis atindividual teacher level. Performance evaluations are often undertaken byschool head teachers or (trained) external observers and financial rewards areallocated on perceptions of teaching quality.2. Performance related pay schemes reward individual teacher performancebased on objective measures of pupil performance. This includes specificindicators such as measurements of pupil progress and pupil performance inend-of-year examinations.303. Finally, school level incentive programmes distribute bonuses to schools onthe basis of performance measured at school level. Depending on the scheme,school performance can be measured in absolute terms - often against aperformance threshold - or relative to other schools (although the latterrequires adjustments to account for differences between schools).There is strong evidence that teachers do respond to financialincentives. Several studies suggest that this response does notuniversally affect all students: most of the improvement appears tocome from previously weak students performing better under suchschemes.All studies suggest that directly rewarded outcomes improve underschool and teacher level incentive schemes. The evidence onunrewarded outcomes is, however, inconclusive.One study suggests that subjective, teacher-based merit pay schemestend to be much more effective if rewards are distributed selectively.Another study comparing the impact of a group versus an individuallevel performance pay scheme finds that the individual level scheme ismore cost effective.Nearly all studies evaluate a sample of schools which areunrepresentative or specially selected to take part. This means that caremust be taken in applying the results of studies more generally.From a welfare perspective, there is still little understanding of theprocesses of change within schools that have adopted a financialincentive scheme, and subsequent effects on staff morale. Neither is

there an understanding of the full costs and benefits of these schemes.US evidenceOwing to a substantial amount of freedom over compensation policy at State level,there are US examples for all three types of teacher incentive scheme. In a crosssectionalstudy, Figlio and Kenny (2006) examine the relationship between school31performance and the presence of a merit pay scheme for 502 schools across the US.Using a rich dataset of school and pupil characteristics to net out any pupil basedvariation between schools, they are able to associate the presence of one or moremerit pay schemes at a school with a 1.3 to 2.1 point increase in test scores. While thisis a small effect in absolute terms, Figlio and Kenny’s analysis suggests that thepresence of a merit pay scheme has three times the impact that maternal education hason pupil performance (a widely recognised determinant of education). Importantly,they obtain this result by narrowing their definition of a merit pay scheme to moreselective programmes only. Indeed, when they expand their definition to include bothselective and non-selective merit pay schemes, they do not observe any significantimpact. This suggests that only relatively selective merit pay schemes are correlatedwith higher pupil performance. Unfortunately, the cross-sectional nature of theirdataset prevents Figlio and Kenny from proposing anything more than associationsand correlations.One of the most sophisticated US programmes was Tennessee’s CareerLadder evaluation system, which rewarded performance, based upon a range ofsources and evaluations, with progress up a rigid career ladder. Each new level wasassociated with salary rises and professional development opportunities. Achievinglevel I would have added $1000, while a level III teacher would have received asalary supplement of $7000. Dee and Keys (2004) evaluate the Tennessee merit payscheme at the same time as a separate experiment was randomly allocating pupils andteachers to their class types. As the assignment of students and teachers to classes wasrandom, this should eliminate any significant correlation between teacher type andstudent ability which would invalidate Dee and Key’s results. The authors comparethe performance of pupils in classes taught by merit pay-based teachers with thosetaught by teachers not participating in the scheme.Using data from a representative selection of 79 state schools over 4 yearsfrom 1985, Dee and Keys (2004) find tha t pupils taught by career ladder teachers had3% higher math scores than pupils taught by non participating teachers. Compared toother factors affecting pupil performance, this increase is equivalent to between 40%and 60% of the gains associated with small class sizes. Their results for math were32statistically significant, but this was not the case for reading scores even though theywere 2% higher for merit scheme teachers. Further analysis shows that higher pupilachievement in math was particularly associated with merit pay teachers on the lowerranks of the Tennessee career ladder. Dee and Keys conclude that while the merit payscheme did significantly improve performance in maths, it is often opposed by teacherunions because of the difficulty in accurately assessing teacher quality.Eberts, Hollenbeck and Stone (2002) compare the outcomes of a mixedmerit/pay for performance scheme implemented at one Michigan high school toanother school with similar characteristics which maintained a fixed compensationpolicy. The mixed programme paid a 12%-12.5% bonus quarterly to teachers whosuccessfully retained more than 80% of students at the end of the quarter.Furthermore, on the basis of very high performance (averaging 4.65/5) in student

evaluations of teaching quality over the year, teachers could become eligible for a 5%salary supplement and an extra 10% increase in their retention bonus.The authors examine changes in attendance rates and exam performance usingdata from both the incentive school and the non- incentive school before and after themixed programme was initiated. In the school which implemented the mixedincentive scheme, Eberts, Hollenbeck and Stone find a 40% increase in retention ratesafter the programme was introduced. However, for the incentive school, they observea 0.53 point drop in exam performance over the whole period, which is greater thanthe 0.37 point decline measured in the non incentive school. This suggests that theincentive programme resulted in the retention of more low-achieving students. Thiskind of response is, therefore, strong evidence that teachers do respond to directlyrewarded performance measures, but it is also a warning that the effects of unintendedconsequences may mean that even closely related outcomes can suffer if not rewardeddirectly.A US example of a group level incentive scheme comes from Dallas, Texas.The Dallas scheme issued bonuses to the top 20% of schools ranked on the basis ofpupil improvement. Compared to some other group level incentive programmes, theDallas scheme was relatively sophisticated:33Choosing a wide range of objective performance measures including drop outrates limited the scope for teachers to concentrate all their efforts on a singleexam.The programme tried to reward effort equally by netting out many socioeconomicfactors that could affect pupil’s learning ability.Once these other factors have been taken into account, schools were rankedaccording to the overall improvement that their students had made across allsubjects, compared to the previous year.Dallas issued $1000 bonuses to each teacher and manager in the top 20% of schoolsand $500 bonuses to each manual worker. The scheme also paid a $2000 activitybudget supplement to the school itself.Ladd (1999) measures the impact of the scheme from its inception in 1991 forfour years, looking specifically at changes in the math and reading pass rates at Dallasschools compared to similar schools in nearby cities. Pass rates are a relatively strongmeasure of student outcomes because school performance was initially very poor.Compared to pass rates in other unreformed city schools, Ladd finds readingoutcomes improve between 10% and 15% more in Dallas schools over the 4 yearperiod, while for maths, the improvement ranges from 14% to 17%. Interestingly,results appear to rise over time. Ladd then examines the pupil performance of specificracial groups and finds that nearly all the improvement comes from Hispanic andwhite student scores. Black student scores do no t show any significant response to theincentive scheme, and Ladd’s results are unable to identify any reasons for this.Evidence from IsraelA similar school level scheme that, like the Dallas programme, rewarded relativeimprovements in student performance was launched in some 62 religious and secularIsraeli schools in 1995. The scheme distributed a reward budget of $1.4m to the topthird of schools in rank order. Once again, performance was measured by a widerange of criteria including dropout rates. In contrast to the Dallas scheme, however, aschool’s rank determined the absolute size of the bonus awarded. Once awarded, thescheme required successful schools to distribute 75% of their bonus to all teaching34

staff proportionate to their salaries, while the other 25% remained with the school. Asan example of the size of rewards available, bonuses in the 1997 year ranged from$13,250 to $105,000 per school. Using a detailed dataset, Lavy (2002) measures theimpact of the scheme on pupil achievement and also compares this scheme to aresources-based programme that provided 22 schools with $1.2m of additionalteaching resources.Lavy finds evidence of significantly higher student achievement in secularschools two years after the programme was introduced, while the results fromreligious schools suggest a positive and significant impact after just one year. Indeedreligious schools appear to have responded better to incentives in both years. Twoyears after the programme was introduced, average subject scores improved by 1.75points for secular schools but by 3.1 points for religious schools. Dropout rates, alsoincluded in the incentive scheme, fell by 0.5% after two years for both secular andreligious schools, reaching 5.5% and 3.2% respectively in 1997. Lavy next examinesthe compositional background of students who demonstrate the greatest improvementafter the incentive scheme was introduced. He found that while the group incentiveprogramme had little overall effect on the proportion of students passing the keymatriculation certificate, those from poorer families observed a large performanceimprovement. Finally, compared to the incentive scheme, Lavy estimates that theimpact of the resources-base programme was 50-70% greater in magnitude for creditunits achieved, average pupil scores and the proportion of students taking the finalmatriculation exam. However, on other measures including science credits achieved,dropout rates and matriculation pass rates, he found that the resources programmeelicited no significant response, in contrast to the incentive programme. Weighing upthe impact of both schemes, their relative costs, and the number of schools involved,Lavy declares the incentive scheme to be more cost effective.Having identified a relatively successful group level incentive scheme, Lavy(2004) moves on to evaluate a teacher level incentive programme. 48 Israeli highschools participated, selected on the basis of previous poor performance. In this case,the size of reward for each teacher depended upon their class’s improvement inexams. After netting out various socioeconomic pupil characteristics, all 207 English35teachers and 237 maths teachers taking part were ranked according to their classscores. Teachers with positive scores received a bonus, with successively higherquartiles eligible for larger financial bonuses. Teachers in the highest quartile receiveda bonus worth 25% of average teacher salary. To reduce the potential for strategicbehaviour, students who did not take the matriculation exams were given scores ofzero.Lavy compares the performance of similar (propensity matched) students innon-incentive and incentive classes before and after the scheme was introduced in2001. He observes an 18% (for maths) and a 10% (for English) increase in the numberof exams attempted by students who were being taught by incentive teachers. Passesin these exams are vital to obtain a place at university in Israel. The programmeresulted in 5.4% more students passing the pre-universit y matriculation exams inmaths while the improvement for English was 4.2%. Breaking down the results forability, Lavy finds nearly all of the impact of the scheme occurred within the lowertwo ability quartiles, the below average students. There is also some evidence that thescheme had a positive impact on the performance of students in subjects withoutincentives attached. For both history and biology (which were not linked toincentives), students appear to have taken more exams, but the pass rate has remained

steady.Looking specifically at the behavioural response of teachers to incentives, theresults of a survey of maths and English teachers suggests that teachers increased afterhours, unpaid help. In other analysis, tests intended to identify strategic behaviour byteachers manipulating the system or cheating, produce no conclusive evidence.Finally, Lavy estimates that this teacher level scheme was approximately twice aseffective as the group level scheme evaluated in Lavy (2002), while costing $100 lessper student.UK evidenceIn 1999, the UK government initiated a teacher level merit pay scheme for stateschool teachers. In some respects, the UK programme resembles the TennesseeCareer Ladder programme evaluated in Dee and Keys (2004). Here, all teachers onthe highest salary level were eligible, equal to approximately 75% of teachers.36Participants must pass an initial performance appraisal carried out by the schoolprincipal, during which both subjective and objective measures of performance areconsidered. Success at this stage was rewarded by a permanent £2000 salary increaseand the opportunity to raise their salary further by passing additional performancereviews. Funded by a separate central government fund with no quotas or limits,success rates for the initial bonus amongst eligible teachers were very high in its firstyear (97% nationally). However, subsequent levels have introduced greater payvariation.Atkinson et al. (2004) describe the details of the first introduction and evaluateits impact on an objective measure of pupil progress, “Value Added”. Value Addedshows the average improvement in learning outcomes for each teacher based on theimprovement of their students in national exams taken at 14 and 16. Atkinson et al.(2004) follow 182 teachers teaching either maths, English or science at one of 18 UKschools. The sample has some differences from the average UK school due tonecessary data demands. In an unexpected initial finding, they observe a substantialamount of variation in Value Added scores obtained by each teacher from year toyear, almost comparable to average degree of variation between teachers. Theiranalysis, which attempts to net out any outside factors that affect student’s learningability, suggests that on average, the pupil of a merit pay teacher gains an additionalhalf grade of Value Added above what is achieved by pupils of non merit payteachers. They also find that most of the improvement in pupil progress is beingachieved by below average students, a result that holds even when high achievers whoonly score the top grades (and hence hit an achievement “ceiling”) are removed fromthe analysis. Finally, examining class assignment of pupils over time, they observe noevidence of merit pay teachers strategically picking high performing pupils.Besley and Machin (2006) looks at the pay of UK head teachers and theperformance of the schools they are associated with. They find that the market forhead teachers creates a pay premium for effective leaders, which, as a directconsequence, imparts performance incentives on all head teachers.37

ConclusionWe find evidence that financial incentives can yield productivity improvements forsome public sector workers. The evidence is strongest for civil servants and teachersbut relatively weak for healthcare workers. Perhaps the main reason for thepredominance of team based incentive schemes in the public sector is that employeesmay be concerned about the effects that individual level schemes could have on

worker morale. Only one study has sufficient information to conclude that workersare less responsive to a team-based pay for performance programme than a schemewhich rewards individual performance. There are also only a handful of studies whichare able to suggest that smaller teams do appear to be more responsive to incentivesthan larger teams. Many of the individual level programmes implemented so far haverewarded performance relative to peers both inside and outside the immediate group.This has perhaps helped to limit any impact that such schemes might have on workermorale.Some studies of civil servants and healthcare workers have found evidence ofunintended consequences and strategic behaviour by workers. While there is evidencefrom one study that strategic behaviour does appear to impose real costs on society,the prevalence of this kind of behaviour may be more a reflection of bad programmedesign than a problem with financial incentives in the public sector per se. Using amixture of subjective and objective performance schemes might help to reducestrategic behaviour in circumstances when output is difficult to measure.Although some studies do attempt to make comparative value for money calculations,there are as yet no UK estimates of the full welfare cost of any incentive schemes inthe public sector. One reason for this is the difficulty of measuring the impact of suchschemes on recruitment. Another reason is that the period of analysis of nearly allpublic sector incentive schemes was too short: few schemes were in place for longerthan a year.Finally, while evidence from the UK public sector is limited, results from theUK generally support the conclusions of international studies. Nevertheless, care mustbe taken when generalising the results of individual studies across other countries orindustries. Most empirical studies have typically relied on non-representative samples38for their results. Indeed, the individual nature of each scheme is important becauseincentive schemes should reflect the environment in which they are operating. Smalldesign differences can have important consequences for the behaviour of employees,and all financial incentive schemes need to be carefully adapted to prioritise localneeds.39

ReferencesTheoretical PapersAcemoglu, Daron; Kremer, Michael; Mian, Atif (June 2003) “Incentives in Markets,Firms and Governments” NBER Working Paper 9802http://www.nber.org/papers/W9802Akerlof, George A and Kranton, Rachel E (2005) “Identity and the Economics ofOrganisations” (2005) Journal of Economic Perspectives Volume 19, Number 1,Winter 2005, pp. 9-32(24)http://www.ingentaconnect.com/content/aea/jep/2005/00000019/00000001/art00002Albano, Gian Luigi and Leaver, Clare (December 2004) “Transparency, Recruitmentand Retention in the Public Sector” Mimeo, Research Centre for Economic Learningand Social Evolution http://privatewww.essex.ac.uk/~itbara/pewg/tranpewg.pdfAsch, Beth (1993) “Designing Military Pay Contributions and Implications of theEconomics Literature” RAND Monograph Reporthttp://www.rand.org/pubs/monograph_reports/MR161/Asch, Beth J. and John T. Warner (2001) “A Theory of Compensation and PersonnelPolicy in Hierarchical Organizations with Application to the United States Military”Journal of Labor Economics, vol. 19, no. 3, 523-562

http://www.journals.uchicago.edu/cgi-bin/resolve?JOLE010701PDFAuriol, Emmanuelle; Guido Friebel, and Lambros Pechlivanos (2002) “CareerConcerns in Teams” Journal of Labor Economics, volume 20 pages 289–307http://www.journals.uchicago.edu/cgi-bin/resolve?id=doi:10.1086/338240Beaulieu, Nancy Dean and Dennis R. Horrigan (2005) “Putting Smart Money to Workfor Quality” Improvement Health Services Research 40 (5p1), 1318–1334http://www.blackwell-synergy.com/doi/abs/10.1111/j.1475-6773.2005.00414.xBenabou, Roland and Tirole, Jean (2003) “Intrinsic and Extrinsic Motivation” Reviewof Economic Studies Vol. 70, 489-520 http://www.blackwellsynergy.com/links/doi/10.1111%2F1467-937X.00253Besley, Timothy and Ghatak, Maitreesh (2003) “Incentives, Choice AndAccountability In The Provision Of Public Services” Institute of Fiscal StudiesWorking paper 03/08 http://www.ifs.org.uk/wps/wp0308.pdfBevan, Gwyn and Hood, Christopher. (2006) “What’s Measured Is What Matters:Targets And Gaming In The English Public Health Care System.” Public40Administration 84:3, 517–538 http://www.blackwellsynergy.com/doi/abs/10.1111/j.1467-9299.2006.00600.xCasalino, Lawrence P. and Arthur Elster (2007) “Will Pay-For-Performance AndQuality Reporting Affect Health Care Disparities?” Health Affairs, 26, no. 3 w405-w414 http://content.healthaffairs.org/cgi/content/abstract/26/3/w405Chambers, Robert G., John Quiggin (2005) “Incentives and Standards in AgencyContracts” Journal of Public Economic Theory 7 (2), 201–228 http://www.blackwellsynergy.com/links/doi/10.1111/j.1467-9779.2005.00201.xChoi, Stephen and Gulati, Mitu (Jan 2004) “A Tournament of Judges?” UC BerkeleySchool of Law http://papers.ssrn.com/sol3/papers.cfm?abstract_id=394700Delfgaauw, Josse and Dur, Robert (2006) “Incentives and Workers’ Motivation in thepublic sector” Tinbergen Institute Discussion Paper, Netherlandshttp://ftp.tinbergen.nl/discussionpapers/04060.pdfEggleston, Karen (2005) “Multitasking and mixed systems for provider payment”Journal of Health Economics Volume 24, Issue 1, Pages 211-223http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V8K-4DNRTWC-1&_user=121739&_coverDate=01%2F31%2F2005&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000010018&_version=1&_urlVersion=0&_userid=121739&md5=31e99c2efcdb6576a687faa8f5e7cbc4Glazer, Amihai (2004) “Motivating Devoted Workers” International Journal ofIndustrial Organisation Vol. 22, No. 3 427-440http://www.mgmt.purdue.edu/centers/ijio/Accepted/2210.pdfKaarbøe, Oddvar M., Trond E. Olsen (2006) “Career Concerns, Monetary Incentivesand Job Design” Scandinavian Journal of Economics 108 (2), 299–316.http://www.blackwell-synergy.com/doi/abs/10.1111/j.1467-9442.2006.00453.xKoch, Alexander K. and Eloic Peyrache (May 5, 2006) “Tournaments, IndividualizedContracts and Career Concerns” Mimeo Royal Holloway and HEC, Paris.http://www.bwl2.uni-bonn.de/tourncontr.pdfLazear, Edward P. (2003) “Teacher incentives” Swedish Economic Policy ReviewVol. 10 No. 2, p. 179-214http://www.ekonomiskaradet.se/Panda_ekonomiska/Data/Documents/sepr2003/Lazear.pdfLeaver, Clare (2004) “Bureaucratic Minimal Squawk Behaviour: Theory andEvidence from Regulatory Agencies” Mimeo, Research Centre for Economic

Learning and Social Evolutionhttp://www.nottingham.ac.uk/economics/research/seminars/papers/c.leaversqwk04.pdf #41Posner, Richard A. (1995?) What Do Judges Maximize? (The Same Thing EverbodyElse Does) http://www.law.uchicago.edu/Lawecon/WkngPprs_01-25/15.RAP.Judges.pdfPrendergast, Canice (2007) “The Motivation and Bias of Bureaucrats” AmericanEconomic Review, Vol. 97, No.1, 180-196http://www.ecare.ulb.ac.be/ecare/seminars/2003-2004/papers%202003-2004/cp1912.pdfRatto, Marissa, with Simon Burgess, Bronwyn Croxson, Ian Jewitt and Carol Propper,(June 2001) “Team-Based Incentives in the NHS: An Economic Analysis” CMPOWorking Paper Series No. 01/37http://www.bris.ac.uk/cmpo/workingpapers/wp37.pdfSchumacher, Johanna (September 2004) “A referee’s report on the article‘Competition and Incentives with Motivated Agents’ by Besley and Ghatak” Topicsin Public Economics http://www.nek.lu.se/NEKFRA/Topics/Assign04/besley.pdfStrunk, Bradley C. and Robert E. Hurley (May 2004) “Paying For Quality: HealthPlans Try Carrots Instead Of Sticks” Center for Studying Health System Change IssueBrief No. 82 (Washington DC) http://www.hschange.org/CONTENT/675/675.pdfVollaard, Ben (2003) “Performance contracts for police forces” CPB Document No.31 http://www.cpb.nl/eng/pub/cpbreeksen/document/31/doc31.pdfEmpirical PapersAtkinson, Adele, Simon Burgess, Bronwyn Croxson, Paul Gregg, Carol Propper,Helen Slater, Deborah Wilson (2004) “Evaluating the Impact of Performance-relatedPay for Teachers in England” CMPO Working paper Series No. 04/113http://www.bris.ac.uk/depts/CMPO/workingpapers/wp113.pdfBaicker, Kate and M. Jacobson (2007) Finders Keepers: Forfeiture Laws, Policing,and Local Budgets, forthcoming in Journal of Public Economics, 2007 (also NBERWorking Paper 10484) http://www.nber.org/papers/W10484.pdfBandiera, Oriana; Barankay, Iwan and Rasul, Imran (2005) “Social Preferences AndThe Response To Incentives: Evidence From Personnel Data” The Quarterly JournalOf Economics Vol. 120 No.3, pages 917-962Besley, Timothy and Machin, Steven (2006) “Are Public Sector CEOs Different?Leadership Wages and Performance in Schools” Preliminary Draft - Not to be Quotedwithout the Authors’ Permission, London School of Economics (October 2006)http://www.ksg.harvard.edu/inequality/Seminar/Papers/Machin062.pdf42Bognanno, Michael (2001) “Corporate Tournaments” Journal of Labour Economics,Vo. 19, No. 2 290-316 http://ideas.repec.org/a/ucp/jlabec/v19y2001i2p290-315.htmlBorzaga, Carlo and Tortia, Ermanno (2006) “Worker Motivations, Job Satisfaction,and Loyalty in Public and Nonprofit Social Services” Nonprofit and Voluntary SectorQuarterly, Vol. 35, No. 2, 225-248http://nvs.sagepub.com/cgi/content/abstract/35/2/225Boyne, George A. and Chen, Alex A. (August 9, 2006) “Performance Targets andPublic Service Improvement” Journal of Public Administration Research and Theoryhttp://jpart.oxfordjournals.org/cgi/reprint/mul007v1.pdfBurgess, Simon; Croxson, Bronwyn; Gregg, Paul and Propper, Carol (July 2001)“The Intricacies of the Relationship Between Pay and Performance for Teachers: Do

teachers respond to Performance Related Pay schemes?” CMPO Working PaperSeries No. 01/35Burgess, Simon; Propper, Carol; Ratto, Marisa and Tominey, Emma (2004)“Incentives in the Public Sector: Evidence from a Government Agency” CMPOWorking Paper Series No. 04/103http://www.bris.ac.uk/cmpo/workingpapers/wp103.pdfChen, Qi and Wei Jiang, (June 2005) “Career Concerns and the Optimal Pay-for-Performance Sensitivity”Mimeo, Fukua School of Business, Duke University andColumbia Business school http://www.columbia.edu/~wj2006/career_concerns.pdfChristianson, Jon B. PhD, David J. Knutson BA, Roger S. Mazze PhD (2006)“Physician Pay-For-Performance. Implementation and Research Issues” Journal ofGeneral Internal Medicine 21 (s2), S9–S13 http://www.blackwellsynergy.com/doi/abs/10.1111/j.1525-1497.2006.00356.xCongressional Budget Papers (Dec. 1995) “Military Pay and the Rewards forPerformance” http://www.cbo.gov/ftpdoc.cfm?index=5307Courty, Pascal and Marschke, Gerald (2003) “Dynamics of Performance MeasurmentSystems” Oxford Review of Economic Policy, Vol. 19, No. 2http://oxrep.oupjournals.org/cgi/content/abstract/19/2/268Courty, Pascal and Marschke, Gerald (2004) “An empirical Investigation of GamingResponses to Explicit Performance Incentives” Journal of Labour Economics, Vol.22, no. 1, 23-56 http://www.journals.uchicago.edu/cgibin/resolve?id=doi:10.1086/380402Crewson, P. E. (1997) “Public Service Motivation: Building Evidence of Incidenceand Effect”, Journal of Public Administration Research and Theory Vol.7, 499–519.http://www.jstor.org/view/10531858/ap040028/04a00010/043Cullen, Julie Berry and Reback, Randall (May 2006) “Tinkering toward accolades:School gaming under a performance accountability system” NBER Working Paper12286 http://www.econ.barnard.columbia.edu/working_papers/wp0601.pdfDee, Thomas S. and Benjamin J. Keys (2004) “Does Merit Pay Reward GoodTeachers? Evidence from a Randomized Experiment” Journal of Policy Analysis andManagement, Vol.23, No.3, 471 –488Doran, Tim M.P.H., Catherine Fullwood, Ph.D., Hugh Gravelle, Ph.D., David Reeves,Ph.D., Evangelos Kontopantelis, Ph.D., Urara Hiroeh, Ph.D., and Martin Roland,D.M. “Does Merit Pay Reward Good Teachers? Evidence from a RandomizedExperiment” Journal of Policy Analysis and Management, Volume 23, Issue 3 (p 471-488) http://www3.interscience.wiley.com/cgi-bin/abstract/109072269/ABSTRACTEberts, Randall; Hollenbeck, Kevin and Stone, Joe (2002) “Teacher PerformanceIncentives and Student Outcomes” Journal of Human Resources Vol. 37, No. 4http://www.jstor.org/view/0022166x/sp030007/03x0075w/0Figlio, David N. and Kenny, Lawrence (Issued in October 2006) “Individual TeacherIncentives and Student Performance” NBER Working Paper No. 12627http://www.nber.org/papers/w12627Frank, Richard G. and Meredith B. Rosenthal (2006) “What Is the Empirical Basis forPaying for Quality in Health Care?” Medical Care Research and Review, Vol. 63, No.2, 135-157http://mcr.sagepub.com/cgi/content/abstract/63/2/135?ijkey=9461803f81d9768b92700f98eded99eeb9a9facf&keytype2=tf_ipsecshaFrey, Bruno S and Jegen, Reto (2001) “Motivation Crowding Theory” Journal ofEconomic Surveys, Vol. 15, No. 5, 589-611

http://www.landecon.cam.ac.uk/speer/iewwp026.pdfGauld, Robin (March 2007) ‘Principal-Agent Theory And Organisational Change’,Policy Studies, 28:1, 17 – 34 http://dx.doi.org/10.1080/01442870601121395Gaynor, Martin, James B. Rebitzer, and Lowell J. Taylor (2004) “PhysicianIncentives in Health Maintenance Organizations” Journal of Political Economy,volume 112 (2004), pages 915–931http://www.journals.uchicago.edu/cgibin/resolve?id=doi:10.1086/421172Gielen, Anne C., Marcel J.M. Kerkhofs and Jan C. van Ours (July 2006)“Performance Related Pay and Labor Productivity” IZA Discussion Paper No. 2211http://ftp.iza.org/dp2211.pdfGlewwe, Paul; Ilias, Nauman and Kremer, Michael (November 2002) “TeacherIncentives” Mimeo, Brookings Institution, Washington DC and NBER Working paperNo. 9671 http://www.nber.org/papers/w967144Gravelle, Hugh; Matt Sutton and Ada Ma (2007) “Doctor behaviour under a pay forperformance contract: Evidence from the quality and outcomes framework”University of York: Centre for Health Economics, Research Paper 28http://www.york.ac.uk/inst/che/pdf/rp28.pdfGrossbart, Stephen R. (2006) “What’s the Return? Assessing the Effect of “Pay-for-Performance” Initiatives on the Quality of Care Delivery” Medical Care Research andReview, Vol. 63 No. 1 http://mcr.sagepub.com/cgi/content/refs/63/1_suppl/29SHeckman, J., C. Heinrich, and J. Smith (2002) “The Performance of PerformanceStandards” The Journal of Human Resources 37: 778–811http://links.jstor.org/sici?sici=0022-166X(200223)37%3A4%3C778%3ATPOPS%3E2.0.CO%3B2-DHeinrich, C. (2002) “Outcomes-Based Performance Management in the Public Sector:Implications for Government Accountability and Effectiveness.” PublicAdministration Review 62: 712–725 http://www.blackwellsynergy.com/doi/abs/10.1111/1540-6210.00253Heinrich, C. (2004) “Improving Public-Sector Performance Management: One StepForward, Two Steps Back?” Public Finance and Management 4(3): 317–351http://www.pmranet.org/conferences/georgetownpapers/Heinrich.pdfHeinrich, Carolyn J (2007) “False or Fitting Recognition? The Use of HighPerformance Bonuses in Motivating Organizational Achievements” Journal of PolicyAnalysis and Management Vol. 26 issue 2, 281-304http://www3.interscience.wiley.com/cgi-bin/fulltext/114177540/PDFSTARTJacob, Brian and Levitt, Steven (2003) “Catching Cheating Teachers: The Results ofan Unusual Experiment in Implementing Theory.” Brookings-Wharton Papers onUrban Affairs p.185-209http://pricetheory.uchicago.edu/levitt/Papers/JacobLevittCatchingCheating2003.pdfJensen, Paul H., Robin E. Stonecash (2005) “Incentives and the Efficiency of PublicSector-outsourcing Contracts” Journal of Economic Surveys 19 (5), 767–787.http://www.blackwell-synergy.com/doi/abs/10.1111/j.0950-0804.2005.00267.xJurges, Hendrik; Richter, Wolfram F. and Schneider, Kerstin (31st May 2005)“Teacher quality and incentives: Theoretical and empirical effects of standards onteacher quality” DIW Berlin, CESifo, Munich and IZA, Bonn http://www.mea.unimannheim.de/mea_neu/pages/files/nopage_pubs/pxmghpz02xuj0nav_91-2005.pdfKahn, Charles M; Emilson C. D. Silva, James P. Ziliak (2001) “Performance-basedWages in Tax Collection: The Brazilian Tax Collection Reform and its Effects” TheEconomic Journal Vol. 111 No.468, 188–205 http://www.blackwellsynergy.

com/doi/abs/10.1111/1468-0297.0059445Knot, Ondrej (March 2005) “The Role of a Mission in the Public Sector: The Case ofCzech and English Judiciary” Mimeo, Charles University, Prague https://www.cergeei.cz/pdf/dp/DP140_2005.pdfKosiak, Steven M. (2005) “Military Compensation: Requirements, Trends andOptions” Centre for Strategic and Budgetary Assessments, Washington DC.http://www.csbaonline.org/4Publications/PubLibrary/R.20050218.Personnel/R.20050218.Personnel.pdfLadd, Helen F (1999) The Dallas school accountability and incentive program: anevaluation of its impacts on student outcomes, Economics of Education Review,Elsevier, vol. 18(1), pages 1-16, Februaryhttp://ideas.repec.org/a/eee/ecoedu/v18y1999i1p1-16.htmlLavy, Victor (2002) Evaluating the Effect of Teachers’ Group Performance Incentiveson Pupil Achievement, Journal of Political Economy, University of Chicago Press,vol. 110(6), pages 1286-1317, Decemberhttp://ideas.repec.org/a/ucp/jpolec/v110y2002i6p1286-1317.htmlLavy, Victor (2003) Paying for Performance: The Effect of Teachers’ FinancialIncentives on Students’ Scholastic Outcomes, CEPR Discussion Papers 3862,C.E.P.R. Discussion Papers. http://ideas.repec.org/p/cpr/ceprdp/3862.htmlLavy, Victor (2007) “Using Performance-Based Pay to Improve the Quality ofTeachers” Future of Children Vol. 17 No.1 Spring 2007 Princeton/ Brookingshttp://www.futureofchildren.org/usr_doc/7_05.pdfLavy, Victor (Issued in July 2004) “Performance Pay and Teachers’ Effort,Productivity and Grading Ethics” NBER Working Paper No. 10622http://www.nber.org/papers/10622Lazear, Edward P. (2000) “Performance Pay and Productivity” The AmericanEconomic Review Vol. 90 No.5 p.1346-1361Levin-Scherz, Jeffrey; DeVita, Nicole and Timbie, Justin (2006) “Impact of Pay-for-Performance Contracts and Network Registry on Diabetes and Asthma HEDISMeasures in an Integrated Delivery Network” Medical Care Research and ReviewVol. 63; No. 1, Supplement S14-S28http://mcr.sagepub.com/cgi/content/abstract/63/1_suppl/14SMarsden, David and Belfield, Richard (2006) “Pay for Performance when output ishard to measure: the case for performance pay for school teachers” [online] LSEResearch Online http://eprints.lse.ac.uk/850/Marsden, David and Ray Richardson (1992) “Motivation And Performance RelatedPay In The Public Sector: A Case Study Of The Inland Revenue” Centre for46Economic Performance, London School of Economics, Discussion Paper No.75http://cep.lse.ac.uk/pubs/download/DP0075.pdfMartin G. (1994) “Performance-related Pay in Nursing: Theory, Practice andProspect” Health Manpower Management, Volume 20, Number 5, 1994, pp. 10-17(8)http://www.ingentaconnect.com/content/mcb/039/1994/00000020/00000005/art00002Mas, Alexandre (Issued in May 2006) “Pay, Reference Points, and PolicePerformance” NBER Working Paper No. 12202 http://www.nber.org/papers/w12202McElduff, P, G Lyratzopoulos, R Edwards, R F Heller, P Shekelle and M Roland(2004) “Will changes in primary care improve health outcomes? Modelling theimpact of financial incentives introduced to improve quality of care in the UK”Quality and Safety in Health Care 2004 No.13: 191-197

http://qshc.bmj.com/cgi/content/abstract/13/3/191Metawie, Miral and Gilman, Mark (September 22-23, 2005) “Problems With TheImplementation Of Performance Measurement Systems In The Public Sector WherePerformance Is Linked To Pay: A Literature Review Drawn From The UK”Department of Industrial Relations, University of Kent at Canterburyhttp://www.kent.ac.uk/kbs/crbp/pdf/Paper%20-%20Miral%20Metawie.pdfMizala, Alejandra and Pilar Romaguera (2004) “School And Teacher PerformanceIncentives: The Latin American Experience”, Center for Applied EconomicsDepartment of Industrial Engineering, University of Chile (March 2004) Published inthe international journal of educational developmenthttp://www.webmanager.cl/prontus_cea/cea_2003/site/asocfile/ASOCFILE220030410154027.pdfMullen, Kathleen J., Frank, Richard G., and Rosenthal, Meredith B (April 2007) “CanYou Get What You Pay For? Pay-For-Performance and the Quality of HealthcareProviders” Job Market Paper: Harvard Universityhttp://www.people.fas.harvard.edu/~kmullen/papers/P4P_latest.pdfMuralidharan, Karthik and Venkatesh Sundararaman (2006) “Teacher Incentives inDeveloping Countries: Experimental Evidence from India” Job Market Paper;Department of Economics, Harvard Universityhttp://econ.ucsd.edu/seminars/0607seminars/Muralidharan_paper.pdfPaarsch, Harry J and Shearer, Bruce (2000) “Piece Rates, Fixed Rates and IncentivesEffects: Statistical Evidence from Payroll Records” International Economic ReviewVol. 41 No. 1 p.59-92Pema, Elda (2005?) “Internal Labour Markets Revisited: Tournaments in Academia”Working Paper, Graduate School of Business and Public Policy, Naval PostgraduateSchool, Monterey, CA http://www.msu.edu/~pemaelda/E%20Pema-Tournaments%20in%20Academia.pdf47Petersen, Laura A, MD, MPH; LeChauncy D. Woodard, MD, MPH; Tracy Urech,BA; Christina Daw, MPH; and Supicha Sookanan, MPH (2006) “Does Pay-for-Performance Improve the Quality of Health Care?” Annals of Internal Medicine Vol.145, No. 4 265-272 http://www.annals.org/cgi/reprint/145/4/265Pink, George H., Adalsteinn D. Brown, Melanie L. Studer, Kristin L. Reiter, PeggyLeatt (2006) “Pay-for-Performance in Publicly Financed Healthcare: SomeInternational Experience and Considerations for Canada” Healthcare Papers, 6(4)2006: 8-26http://www.longwoods.com/product.php?printable=Y&productid=18260&page=2Prendergast, Canice (2001) “Selection and Oversight in the public sector, with the LosAngeles Police Department as an example” NBER Working Paper No. 8664http://papers.ssrn.com/sol3/papers.cfm?abstract_id=294102Prendergast, Canice (1999) “The Provision of Incentives in Firms” Journal ofEconomic Literature Vol.37, No.1, p.7-63Propper, Carol and Wilson, Deborah (2003) “The Use and Usefulness of Performancemeasures in the Public Sector” Oxford Review of Economic Policy Vol. 19 No. 2,250-268 http://oxrep.oxfordjournals.org/cgi/content/abstract/19/2/250Ramseyer, J Mark and Rasmusen, Eric B (1997) “Judicial Independence in a CivilLaw Regime: The evidence from Japan” Journal of Law, Economics and OrganizationVol 13, No. 2 http://jleo.oxfordjournals.org/cgi/content/abstract/13/2/259Rosenthal Meredith B, Frank Richard G, Li Zhonghe and Epstein, Arnold M (2005)“Early Experience with Pay for Performance: from concept to practice” Journal of the

American Medical Association, Vol. 294, No. 14, p 1788-1793 http://jama.amaassn.org/cgi/content/abstract/294/14/1788PriceWaterhouseCoopers (October 2006) “Connecting Public Sector Pay to ServiceDelivery” Government and Public Sectorhttp://www.pwc.com/extweb/pwcpublications.nsf/docid/16F22C47349E066280257218004FCC77Roski, Joachim Ph.D., M.P.H., Robert Jeddeloh M.D.b, Larry An M.D.c, HarryLando Ph.D.d, Peter Hannan M.Stat.d, Carmen Hall M.S.e and Shu-Hong Zhu Ph.D(2003) “The impact of financial incentives and a patient registry on preventive carequality: increasing provider adherence to evidence-based smoking cessation practiceguidelines” Preventive Medicine Volume 36, Issue 3, March 2003, Pages 291-299http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6WPG-47XSXB3-3&_user=121739&_coverDate=03%2F31%2F2003&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000010018&_version=1&_urlVersion=0&_userid=121739&md5=966f66f0ca82610ec1d8f50404c3cdc948Serneels, Pieter; Lindelöw, Magnus; Garcia-Montalvo, Jose and Barr, Abigail (2005)“For Public Service or Money: Understanding Geographical Imbalances in the HealthWorkforce” World Bank Policy Research Working Paper 3686, August 2005http://heapol.oxfordjournals.org/cgi/content/abstract/22/3/128Shearer, Bruce (2004) “Piece Rates, Fixed Wages and Incentives: Evidence from aField Experiment” The Review Of Economic Studies Vol.71, p.513-534Shi, Lan (Jan, 30, 2006) “Does Oversight Reduce Policing? Evidence from theCincinnati” Department of Economics, University of Washingtonhttp://faculty.washington.edu/lanshi/Research/policing_jan_06_lan_shi.pdfStoddard, Christiana and Kuhn, Peter (2006) “Incentives and Effort in the PublicSector: Have U.S. Education Reforms Increased Teachers’ Work Hours?” NBERWorking Paper No. 11970, Issued in January 2006http://www.nber.org/papers/W11970Sutton, Matt, and Gary McLean (2006) “Determinants of primary medical carequality measured under the new UK contract: cross sectional study” British MedicalJournal 332: 389-390 http://www.bmj.com/cgi/content/abstract/332/7538/389Vollaard, Ben (2003) “Performance contracts for police forces,” CPB Documents 31,CPB Netherlands Bureau for Economic Policy Analysis.http://ideas.repec.org/p/cpb/docmnt/31.htmlVyrastekova, Jana; Onderstal, Sander and Koning, Pierre (2006) “Team Incentives inPublic Organisations: An experimental study” CPB Discussion Paper, Netherlandshttp://ideas.repec.org/p/cpb/discus/60.html

THE ECONOMIC APPROACH TO PERSONNEL RESEARCHMichael GibbsGraduate School of BusinessUniversity of Chicago

Chicago, IL 60637&Alec LevensonCenter for Effective OrganizationsMarshall School of BusinessUniversity of Southern CaliforniaLos Angeles, CA 90089We compare the economic approach to research on personnel and organizational designto approaches from behavioral disciplines. Instead of a survey of the field, our emphasisis on topics that are important in organizational research outside of economics, yet havebeen little emphasized by economics. We contend that many of these topics hold greatpromise for insights from the economic approach. In some cases we sketch ways inwhich economists can approach these topics. We also briefly discuss empirical methodsin personnel economics.Forthcoming in The Expansion of Economics,NY: M.E. Sharpe, ed. by Shoshana Grossbard-Shechtman.Prepared for the 1999 meetings of the Society for the Advancement of Behavioral Economics. We have benefited greatly from interactionswith teachers, colleagues, and students over the years. A partial list of those who we wish to thank includes George Baker, GaryBecker, Mike Beer, Susan Cohen, Joe Cooper, David Finegold, Ray Friedman, Cristina Gibson, Richard Hackman, Wally Hendricks,Bengt Holmstrom, Kathryn Ierulli, Mike Jensen, Ed Lawler, Eddie Lazear, Gary Loveman, Bentley MacLeod, Janice McCormick, SueMohrman, Kevin J. Murphy, Canice Prendergast, Dan Raff, Sherwin Rosen, Wim van der Stede, and Karen Wruck.“Definition of an Economist:A person who knows all the answers, but doesn’t understand the questions.”(The Humorous Dictionary of Economics, 1983)“An economist (John) loses his keys while walking across a dark parking lot. His friend (Jane) happens uponhim some time later.Jane: ‘Hi John! What are you looking for?’John: ‘I lost my keys somewhere in this parking lot.’Jane: ‘Did you lose them under one of these streetlights?’John: ‘No, I checked those areas twice already.’Jane: ‘So why are you still looking under the streetlights?’John: ‘Because that’s where the light is!’”IntroductionThe economic approach to personnel and organizations has grown greatly in scope and importance overthe last decade or two. It is now recognized as a separate field within labor economics, responsible for as muchas a third of papers in leading labor journals.1 Business schools increasingly offer personnel economics courses,and hire economists to teach human resource management and other organizational courses traditionally taughtby non-economists. Textbooks using, or strongly influenced by, the economic approach to organizations haveappeared regularly for several years (e.g., Baron and Kreps 1999; Brickley, Smith and Zimmerman 2001; Jensen1998; Lazear 1998; Milgrom and Roberts 1992). Personnel and organizational economics has developed to sucha great extent that there are now many excellent literature reviews that take stock of the contributions or criticizethe approach (see Appendix A for a partial list).

A cynic might worry that the abundance of literature reviews indicates the imminent death of the field.At the very least, many observers criticize the economic approach to personnel. The criticisms often are thoseapplied to all of economics, such as the assumption of rational behavior, crude specifications of individual andgroup psychology (Kaufman (1999) is the most sophisticated critique along these lines), and too-simple modelsof complex reality (Hirsch, Michaels, and Friedman 1987). Another common criticism is that the field focusesnarrowly on incentives to the exclusion of other interesting and important topics.Our purpose is not to provide another survey of the field. Nor is it to join in what Winship and Rosen(1988) term “tiresome debates” about methodology. Instead, we try to be constructive in other ways. One barrierto interdisciplinary communication is differences in language (Merchant, Van der Stede and Zheng 2000). Weuse the language and viewpoints of both perspectives to try to decrease the language barrier. In the first sectionbelow, we briefly outline a way that many scholars outside of economics tend to view organizational design (seeFigure 1): the “systems” view of an organization. We interpret this view in economic language, and use it as theskeleton for the rest of the chapter.Our most important goal is to discuss research areas that we believe hold great promise for gains fromintegrating economic and behavioral approaches. In the spirit of the two quotes at the beginning of the chapter,economics is stronger at providing theory based on mathematical models, and pursuing topics that existing theoryor databases most readily lend themselves to. Organizational scholars outside of economics give more emphasisto topics that are of empirical and practical importance, and build theories accordingly, even where suchtheories do not lend themselves to formal (especially mathematical) modeling. We identify topics that have receivedgreat attention outside economics, little inside economics, and which might benefit greatly from the economicapproach.2We do not provide a comprehensive review of topics in the management literature, with the objective ofdetailing how an economic approach can improve them, for two reasons. First, introducing the topics foreconomists is task enough for a single paper. Indeed, in many cases we give short shrift to topics that have receivedextensive coverage by behavioral researchers. Second, and perhaps more importantly, because many ofthe topics discussed here have been barely researched by economists, it is too soon to predict how taking aneconomic approach will improve the existing knowledge base. That said, we are confident that economics hasmuch to add, as we point out below, and look forward to a day in the near future when such a review will drawon a large body of research completed by economists.The plan of the chapter is as follows. In the next section, we outline the systems view of organizational

design, using it to compare economic and behavioral approaches. This also serves to bring out the main topicswe pursue. In the third section we consider important organizational policies that behavioral scholars have studiedextensively, but that economists have paid little attention to. In that section, our goal is to suggest interestingareas of research for economists to pursue, as well as to sketch some ideas about how that research might bepursued. Topics we discuss include intrinsic motivation, job design, decision-making, organizational structure,and coordination. The fourth section takes a similar approach to topics in organizational dynamics. The fifth sectionprovides a brief discussion of methodological and empirical issues, and the sixth section ends with briefconcluding remarks.The Systems View of Organizational DesignThere are many ways to compare economic and behavioral approaches to organizational design. Herewe highlight one difference that has not often been emphasized: behavioral approaches tend to take more of asystems view than does the economic approach (e.g., Galbraith 1977, Senge 1990). Figure 1 illustrates the systemsview of organizational design. It is adapted from a similar figure in Beer, et al (1984, Fig. 2-1). Similarideas can be found in other writings throughout organizational research for many years. We modified the Beer,et al figure to make it more consistent with the language of economics.“Organizational Design” represents firm design and personnel policies; these are the focus of personneleconomics.2 However, the systems view represents it as part of a larger process. By doing so, it encouragesthinking about broader questions than by focusing only inside the Organizational Design box.Behavioral scholars often view firms by first starting with the “Strategy” and “Environment.” The firmchooses a strategy, product line, and product attributes (emphasis on cost, quality, timely delivery, customer service,innovation, etc.). This depends on the firm’s constraints, including product and labor market conditions,legal considerations, social or political pressures, and competitor strategies. The strategy determines what mustbe done inside the firm in order to produce the product and implement the strategy, which then determines optimalorganizational policies.Organizational policies, in turn, produce “Intermediate Outcomes,” which we have described in termsfamiliar to economists: investment, matching, and motivation. Personnel policies result in investments in firmspecificand general human capital, employee expectations, corporate culture, etc. They result in matching betweenemployee skills and tasks, information and decisions, etc. They result in intrinsic and extrinsic motivation.These outcomes, combined with other firm inputs like capital, result in “Business Outcomes.” BusinessOutcomes are measured in the same way as Strategy objectives: profit, product attributes, innovation, etc. Ofcourse, this simplistic description makes the relationships seem more linear/causal than they really are. Intermediate

and business outcomes interact in complex ways, including feedback from business outcomes to intermediateoutcomes.External and Internal FitOne thing the systems view highlights is that the firm’s design should be complementary, or have externalfit, with the firm’s objectives and constraints. In other words, the Business Outcomes produced by the OrganizationalDesign should ideally match the Strategy. Thus, in behavioral research, an interesting question is3how personnel policies should vary with dimensions of the firm’s product, technology, or strategy. By contrast,virtually all economic models have a single “product” with no dimension other than quantity (and sometimesquality). In an economic model, it typically is irrelevant whether the firm is producing a simple, low-tech, standardizedproduct such as a tin can, or a complex, evolving product such as a semiconductor. It is irrelevantwhether the firm is in manufacturing or white-collar work. In behavioral research, such distinctions are fundamental,and lead to very different insights about optimal policies. We argue below that such distinctions maylead to important insights in personnel economics.This is our first difference between the two approaches: economics tends to be vague about what thefirm is trying to do beyond the abstract notion of “profit maximization.” Because of this, economists have littleto say about how policies should vary with different types of products, competitive environments, or technology,and how they might change over time.The systems view also emphasizes internal fit, the idea that personnel policies may be more effective ifthey are designed to be mutually reinforcing (that is, that they should be viewed as a system). For example, promotion-from-within is often said to be complementary with lifetime employment and deferred compensation.Team production and employee “empowerment” (decentralization of some decision rights) are often said to becomplementary with enriched jobs. In Figure 1, this is the idea that the four types of policies in the OrganizationalDesign box might be designed to work well together. At one level, this is trivial and obvious. At another,though, it may imply that there are patterns of organizational policies that firms tend to (and should) use together(Galbraith 1977; MacDuffie 1995; Lawler, et al. 1998; Baron and Kreps 1999).Economists have recently shown interest in internal fit, drawing on the complementarities work of Milgromand Roberts (1995). For example, Ichniowski and Shaw (1995) and Ichniowski, Prennushi and Shaw(1997) investigated complementarity of personnel policies in steel finishing lines. In fact, their work was inspiredby the behavioral literature’s emphasis on internal fit among personnel policies. Gibbons and Waldman(1999) emphasized the value of integrative models designed to address patterns of empirical evidence. In orderto do so, such models consider a broader set of policies as a system. We believe internal fit is an interesting ideathat deserves more theoretical and empirical analysis.Organizational Design

In the Organizational Design box in Figure 1, the middle two types of policies are by far the main focusof personnel economics. “Internal Labor Markets” involve recruiting, developing, assigning, and retaining theright worker skills. “Performance Evaluation and Incentives” are, of course, the essence of personnel economics.The economic approach has made substantial contributions to our understanding of these policies, and to describingthe links between them and the three kinds of Intermediate Outcomes. If anything, economics has acomparative advantage here. It is the other two areas where economics has had less impact, and economists haveplaced less emphasis.“Job Design and Decision Making” encompasses the definition of jobs (bundling of tasks), teams, andinformation development and processing. It also includes macro organizational structure, such as hierarchy, divisionalization,and coordination mechanisms. We make two observations here. First, the initial work in personneleconomics, the first chapter of Adam Smith’s The Wealth of Nations (1776), is on specialization in job design.However, there has been almost no work on job design in economics ever since. Second, much of the economicwork on organizational structure has focused on the boundaries of the firm (Williamson 1975), ratherthan on what happens inside the boundaries. In contrast, job design and teams are large areas of interest amongbehavioral researchers. This research is often tied to discussions of “intrinsic” motivation, or ways in which jobsmay be designed to make the work more motivating or less onerous. Decision making and organizational structurealso have received much attention outside of economics, though less so than job design issues. We arguebelow that these are perhaps the most promising areas for future work in personnel economics.A second major difference between the two research approaches is that personnel economics tends to bevague about what employees do. In economic models of assignment, training, or incentives, it is usually irrelevantto distinguish between blue-collar and white-collar jobs.3 Workers are employed in generic “jobs” with ab4

stract production functions. The only distinction between a manager and a worker is who is evaluating whom.Teams tend to be modeled as a group of workers with a common performance measure, but the reason for usingteams in the first place is usually ignored. There is little analysis of different forms of hierarchy, coordination, orcollaboration. All of these issues are fundamental to behavioral organizational research.The last type of policy we call “Implicit Contracting.” These practices go by many names, such as “psychologicalcontracting,” “employee voice,” “culture,” and “relational contracting.”4 We mean practices firms useto facilitate implicit agreements with employees, for a variety of reasons, all arising from the inability to writeperfect explicit contracts. A classic example is developing mutual trust to share investments in, and returns from,firm-specific human capital. Another is setting the implicit terms of the employment relationship between thefirm and the employee at hiring. Another is investing in a reputation for fair treatment of employees, to facilitate

subjective performance evaluations. These practices have received great attention outside, and some attentioninside, of economics. The most common method for economic analysis of implicit contracting is game theory,with an emphasis on reputation and repeated games (Kreps 1990). Most of the work inside economics is theoretical,but most outside economics is empirical.Organizational DynamicsThe final message in the systems view in Figure 1 is that organizational design is dynamic. An importanttopic in behavioral research on organizations is “managing change.” Imagine that a firm has strong externaland internal fit, so that the personnel policies are complementary, and well adapted to the product, technology,competitive environment, and strategy. Now suppose that something changes in the environment; e.g., a newtechnology is developed that leaves old production methods outdated and relatively inefficient. In terms of Figure1, this means that the Business Outcomes no longer match the Strategy. How does the organizational designchange? Behavioral scholars often argue that organizational design is costly to change. This implies that the optimalchange may involve not only changing the inside of the organization, but also reformulating the company’sstrategy to reflect constraints from the existing organization. This is why there is a feedback loop fromBusiness Outcomes to Strategy in Figure 1. This topic has received little attention from economists.The second sense in which an organization is dynamic is less reactive, and more proactive. It is possibleto conceive of an organization being designed to be adaptable to future, unforeseeable environmental changes.Similarly, some personnel policies may foster continuous improvement in methods, innovation, learning, etc.These topics receive substantial attention outside, but very little inside organizational economics.These two senses of organizational dynamics lead us to our third major difference between behavioraland economic approaches: economics tends to take a static view of organizational design, which makes it difficultto consider issues of organizational change, or designing adaptable organizations.Neglected Areas Inside the Black BoxIn this section, we discuss areas of organizational design that we believe hold promise for future researchby personnel economists. Our first and most important topic is job design and intrinsic motivation. Wecontrast behavioral and economic approaches to job design (job enrichment v. specialization). Combining thetwo perspectives makes the two seemingly very different approaches consistent. We suggest a way economistscan begin modeling intrinsic motivation. This discussion also highlights the potential benefits of considering notonly complementarity between personnel policies (internal fit), but also external fit; we argue that the firm’s optimalapproach to job design depends importantly on the nature of its product, technology, and environment.Job Design and Intrinsic MotivationIn the first chapter of The Wealth of Nations, Adam Smith analyzes specialization in job design in a pin

factory.5 In what may be the first econometric case study in personnel economics, Smith calculates that if aworker does every task involved in making a pin, he can produce less than 20 pins per day. Smith then estimates5that if workers specialize on different tasks, 10 workers can together produce about 48,000 pins per day. Thus, inSmith’s example, specialization through narrow job design increases production by about 24,000 percent. This isan astonishing figure; even with a very large standard error, it indicates the potential power of specialization asan approach to job design. Since then, a brief discussion of specialization (along with comparative advantage)has been a staple of microeconomics textbooks. There has been only a little work on job design in personneleconomics since, usually focusing on specialization (e.g., Rosen 1983, Becker and Murphy 1992).Contrast this with the treatment of job design in behavioral research. The classic works are Hackmanand Lawler (1971), and Hackman and Oldham (1980), widely cited as the most important works on job designand intrinsic motivation. Their approach is the exact opposite of Smith’s: they argue that job enrichment (addingcomplexity or more tasks) generates intrinsic motivation and increases productivity. The Hackman and Oldham[H&O] model of job design (which builds off the foundation provided by Hackman and Lawler) is so importantoutside economics that we describe it briefly here, and focus our discussion around it. Figure 2 reproduces thediagram of their model (1980, p. 83). Lawler (1986) provides an excellent treatment of job enrichment.H&O focus on designing jobs so that workers are more intrinsically motivated. They argue that three“critical psychological states” are required to achieve this: Meaningfulness, Responsibility for Outcomes, andKnowledge of Actual Results. The more of each of these states that the job design generates in the employee, thegreater should be intrinsic motivation. These three states are determined by five elements of job design: SkillVariety, Task Identity, Task Significance, Autonomy, and Feedback. Finally, three “moderators” (what economistswould call parameters) affect the strength of these effects: Growth Need Strength, Knowledge and Skill,and Context Satisfaction. Threaded through this section, we explain these terms as we interpret the model aseconomists.The most important element of the H&O model is making work “meaningful” so that workers becomemore “involved” in their jobs: are more interested, pay more attention, think more carefully, and work more diligently.The key aspect of job design that generates such involvement in work is Skill Variety: “the degree towhich a job requires a variety of different activities in carrying out the work, involving the use of a number ofdifferent skills and talents of the person” (1980, p. 78). H&O argue that skill variety causes employees to “challengeor stretch their skills or abilities” (p. 78). They go on to say that this is a fundamental characteristic of humanpsychology, using the metaphor of newborn babies who are “wired in” to “explore and manipulate their

environments” (p. 78). In this view, work is more interesting (less onerous) if it is more intellectually challenging.For this reason, they argue that job enrichment increases intrinsic motivation. Indeed, even Smith recognizedin The Wealth of Nations that specialization could lead to boredom and reduced productivity of workers.Job enrichment is important empirically, for at least three reasons. First, there are observable patterns ofspecialization and job enrichment that are ripe for theoretical explanation. For example, management, “knowledgework,” and jobs with higher levels of responsibility clearly tend to be defined with broader sets of tasks.Manufacturing jobs were historically defined with a narrower set of tasks. Second, and perhaps most interestingly,there appears to have been a recent trend toward job enrichment. Third, job enrichment is often associatedwith other policies such as job rotation, team production, and “empowerment.” This is suggestive of the idea ofcomplementarities or internal fit, and as such has attracted some economic researchers. However, this area haslacked a strong theoretical foundation for empirical work (Cappelli and Neumark 1999).Given the importance of job enrichment in behavioral research and in practice, and the contrast with theeconomics focus on specialization, we believe that it is important for personnel economists to seriously explorethis idea. We now outline how we think economists could get a handle on this topic. By doing so, we present adifferent view of the H&O model that we hope illustrates the potential benefits of integrating economic and behavioralapproaches.6Modeling Intrinsic MotivationEconomists tend to avoid assuming things about the utility function, such as that certain types of workdesign raise or lower the onerousness of effort, unless this can be done in a way that produces empirically testableimplications. Therefore, most economic work on intrinsic motivation sidesteps this issue in one way or an6

other. One recent approach avoids the issue by assuming that a worker’s actions affect both firm profits, and theemployee’s utility; that is, the employee gets intrinsic motivation from the output (Murdock 1998).7

Kreps(1997) focuses on how, in a multitask framework, intrinsic motivation on some tasks can be affected by extrinsicincentives on other tasks. Kreps does say, however, that perhaps it is time for economists to try to model intrinsicmotivation more directly.A simple starting point is the observation that job enrichment appears to increase intrinsic motivation.This might be done by assuming that the marginal cost of effort is lower if the worker is assigned more tasks.However, this runs the risk of assuming the result. Moreover, it does not capture the essence of the behavioralliterature, that intellectual “stretch” or “challenge” are important. Therefore, a second possibility is to incorporatesome notion of intellectual challenge about tasks by the worker. There are two senses that seem important.

One is simply learning how to perform new tasks, or how to perform them better. The second is using intellectualor “thinking” skills on the job.For the first sense of how work may be “challenging,” one approach is to incorporate learning by doing(Lindbeck and Snower 2000). Learning would increase the worker’s marginal product of effort, since the workerbecomes more facile at the task. This is the standard economic approach. But the degree of learning by doing,perhaps modeled as the change in the worker’s human capital stock in the task, would also affect (lower) themarginal cost of effort. That would reflect intrinsic motivation through learning.Consider how implications might be derived from such a model. The firm would face a tradeoff: moreskilled workers are more productive, but less intrinsically motivated (holding characteristics of the job constant),because opportunities for further learning are diminished. It would not necessarily be the case that the bestworker for a job is the one who is best trained for the job, since such a worker would have less to learn. This isdifferent from the standard economic view that matching should be by comparative advantage. It is consistentwith the evidence that many organizations using job enrichment also use job rotation: workers are moved to newjobs periodically, even if they have mastered their old jobs. Ordinarily this would seem a puzzle to economists(especially since job rotation reduces gains from specialization even further).H&O discuss two factors affecting intrinsic motivation through job enrichment, the “growth needstrength” moderator and the “knowledge and skill” moderator.8 By growth need strength moderator, they meanthat individuals respond differently to job enrichment. This is equivalent to saying that differences in marginalutility of learning affect how strongly a worker is intrinsically motivated. By knowledge and skill moderator,they mean the extent to which the worker has the requisite skills so that the job will not be too challenging. It isstraightforward to incorporate this into an economic model: the worker’s stock of human capital in the task mayhave non-linear effects on intrinsic motivation. Very low levels of human capital may imply low intrinsic motivationbecause the challenge is too hard; very high levels of human capital may also imply low intrinsic motivationbecause the challenge is too easy. Thus, the optimal matching of worker to job may also imply some minimallevel of training greater than zero.There is a second sense of making the employee more involved in his or her work, beyond learning howto perform the job better: the extent to which the job is intellectually engaging. This is probably best modeled asa property of the work environment, or production function, not the worker’s utility function, although it requiresat least a rudimentary modeling of the idea that the intellectual nature of work may affect intrinsic motivation.Therefore, a second approach that we believe would be fruitful is modeling the information structure andintellectual nature of the production environment.

In addition, it is useful to distinguish between job enrichment and job enlargement. Job enlargement is ahorizontal expansion of tasks without a commensurate increase in the intellectual nature of the job. Job enlargementcan decrease boredom (and errors) by reducing repetition, but does not necessarily increase other aspectsof intrinsic motivation. For example, reducing the number of hours devoted to filing, substituting hours devotedto stuffing envelopes, can enlarge a file clerk’s job. The job can be enriched by including the worker in decisionmaking over what types of documents should be stored electronically versus in paper form.Most economic incentive models use a metaphor that seems to be drawn from physical tasks: the firmwants the worker to provide more “effort.” However, this is not quite what is needed for modeling intellectualwork. In such jobs the firm does not want the employee to work “harder,” but to think more carefully, collect7more relevant information, perform analysis, etc. We encourage theorists to think more carefully about how tomodel these issues. Perhaps most interesting, though, would be for empirical economists to study the relationshipbetween the intellectual nature of the work environment, and the job design chosen by the firm.Which Tasks Should Be Bundled Together?In order to put further content on the idea of job enrichment, it is useful to consider which tasks shouldbe bundled together. The clear answer to an economist is that more complementary tasks (larger positive crossmarginalproducts of production) should be bundled together. Unfortunately, this has little empirical content asstated, since it provides no way of thinking about which tasks are complementary. A second factor in the H&Omodel helps us begin to think about this issue: “Task Identity.” H&O describe Task Identity as “the degree towhich a job requires completion of a ‘whole’ and identifiable piece of work.”An economic interpretation of Task Identity is constructive. H&O seem to be saying that it is importantto modularize production into relatively discrete or separable sub-processes. There are three implications of suchmodularity. First, if production can be modularized, then it must be that the modules capture most of the importanttask complementarities within each module. Indeed, this is almost the definition of modularity. Thus, designingjobs with this concept of Task Identity in mind captures in a practical and implementable way theeconomist’s notion of bundling complementary tasks together. Second, it gives guidance on how far to take jobenrichment. The benefits of job enrichment are limited by the ability of the worker to learn and handle multipletasks on the job. But the benefits are also limited by the nature of the tasks, in that bundling unrelated tasks togetheris less likely to increase productivity.Third, modularity is consistent with our emphasis on designing work to increase learning on the job.When more closely related tasks are bundled together, it seems very likely that the employee will learn moreabout ways to improve production. For example, if an employee produces two parts that work closely together

in a diesel engine, then the employee is likely to make both parts with higher quality, because the worker has abetter understanding of how the parts will be used, how they fit with each other, etc. It also seems plausible thatthe worker will be more likely to figure out process improvements, because the production of each part mayhave implications for the effectiveness of the other part.Where Is Job Enrichment Most Valuable?This perspective helps us to understand trends toward adoption of job enrichment and related techniques.Under the traditional “Taylorist” approach, stemming from the work of Frederick Taylor (1923), industrialengineers worked out the “best” way to do work ex ante. They then designed narrow, specialized jobs,combined with mechanization, to repeat these best practices as much as possible; the goal was to standardizeproduction to wring out variation and defects. This approach makes sense to the extent that the best productionprocess can be figured out ex ante at reasonable cost.9In many environments, figuring out the best approach to job design ex ante will be inefficient for severalreasons. The more complex the product (in terms of number of parts, number of production tasks, or technology),the more difficult is the industrial engineer’s optimization problem. Similarly, more complex product lines,with more variations in product design or more customization, imply costlier industrial engineering. More rapidtechnological change implies that the process must be redesigned more frequently. All imply lower returns toinvesting in ex ante industrial engineering, and thus less ex ante optimization of production methods.The less effective is ex ante industrial engineering, the greater the scope for workers to learn processimprovements. The greater the scope for workers to learn, the more beneficial will be job designs that encouragelearning by workers. This helps explain why job enrichment and related policies are adopted more in some settingsthan in others. There has been a trend toward job enrichment in recent decades partly because firms increasinglyuse customization, and techniques that facilitate more frequent changes in products and product lines(Milgrom and Roberts 1990). It is also likely that more rapid technological change in recent decades has fueledthis trend; thus, such policies should be adopted more in settings where technological change is more important.8Finally, under this view it is not surprising that adoption of job enrichment is closely associated with havingworkers emphasize continuous process improvement and total quality management; both are more important ifthe firm is less able to conduct effective ex ante process optimization.This discussion suggests that economists think about modeling the information properties of the productionenvironment, to try to capture their effects on job enrichment and modularization. For example, theoreticaland empirical researchers might consider incorporating ideas like the following into their models and data collection:how long the firm has been making a given product; how long it anticipates making the product without

major changes in design or production technology; the complexity of the product, product line, or productionprocess; the rate of change in technology in the industry; and whether the production process allows for frequentchanges in products to accommodate customization.Job Enrichment and Decision MakingIf one of the primary benefits of job enrichment is that workers can learn and figure out process improvements,it follows that the firm will want workers to take initiative to some extent: come up with suggestions,and make some decisions about which to try and how to implement them. Two other parts of the Hackman-Oldham model address this.H&O define “Autonomy” as “the degree to which the job provides substantial freedom, independence,and discretion to the individual in scheduling the work and determining the procedures” (p. 79). By Autonomy,H&O do not really argue that workers get intrinsic motivation from being able to work independently (thoughthat might be true). Instead, they emphasize giving the worker responsibility about procedures, in the sense ofbeing able to try new methods (see Fig. 2).“Feedback” is defined as “the degree to which carrying out the work activities .. provides the individualwith direct and clear information about the effectiveness of his or her performance” (p. 80). In other words,workers need to be provided information on the effects of their efforts, if the firm wishes to have workers comeup with initiatives and test them.Thus, decision-making is an important element of job enrichment.10 This explains why employee “empowerment”or “participation” is so strongly associated with job enrichment programs. We can think of the twoas highly complementary practices in the following way. In addition to intrinsic motivation, job enrichmentgives the worker a broad understanding of a set of strongly complementary tasks. The worker gains detailed,potentially sophisticated knowledge about the production process and sometimes the product design. Empowermentgives the worker the ability to make use of this information, trying new methods to improve the processand reduce defects.11 Together they allow the worker and the firm to make improvements and more readily respondto changes in the environment. We now turn to a more detailed discussion of decision-making.Decision MakingPersonnel economists tend to think of decisions as being made within hierarchies. But hierarchy modelsrarely have much element of decision to them; they are usually conglomerations of primitive production functions,with spans of control between supervisors and subordinates (Beckmann 1977; Calvo and Wellicz 1979;Rosen 1982).12 Few models attempt to get a handle on decision-making. Yet decision-making is one of the mostimportant functions of the internal organization of the firm. Location of decision-making authority (the degreeof decentralization) is a major topic in behavioral fields. Moreover, decision making inherently involves processing

and strategically using information, which is a building block of economic theories of organization. It istherefore surprising that economists have had little to say about decision making.Jensen and Meckling (1992), applying Hayek’s (1945) analysis of markets to internal firm organization,provide a starting point. They emphasize a distinction between “specific knowledge,” which is costly to communicateto someone else, and “general knowledge,” which is not costly to communicate. If knowledge is general,decisions can be centralized to ensure coordination. The more that knowledge is specific, the more will the9firm locate decision rights where the specific knowledge is generated. They apply this idea to discuss decentralization,which involves a tradeoff between use of specific knowledge and coordination costs.13

We can enrich the idea of specific knowledge by thinking more specifically about what attributes ofknowledge (information) make it more costly to communicate, and thus push the firm toward decentralized decision-making (Milgrom and Roberts 1992; Brickley, Smith and Zimmerman 1996). One attribute is whether theinformation is perishable. For example, information about a customer’s immediate request for service must beacted upon quickly or it loses value. In such cases the firm is likely to decentralize service decisions, because ofthe time involved in passing the information to a central decision maker.Another attribute of information that makes it costly to communicate is complexity. If information hasmany dimensions (variables), with complex relationships between the variables, it may be costly to describe andconvey to others. This helps to explain why participative decision-making is used in complex work environments.A diesel engine plant is more likely to use job enrichment combined with empowerment than is a tin canfactory: the engine has hundreds of interacting parts, while a tin can has at most half a dozen.A third attribute of information that may make it costly to transfer is whether or not the information iscostly to understand once received. Two examples come to mind immediately: scientific or technical information,and subjective or experiential information. In both cases, decisions that make use of the information willtend to be decentralized to workers who have the knowledge. For example, performance appraisals tend to bedecentralized to supervisors despite problems of favoritism and influence costs (Prendergast 1999), becausemost jobs require subjective evaluations. These are hard to do without actually observing the employee on aday-to-day basis, and costly to communicate.A more subtle issue of the cost of transferring subjective information is that it is more likely to be manipulableby the holder (performance appraisals again come to mind). But that does not necessarily imply thatdecisions should be decentralized to the potential manipulator! On the contrary, the opposite may sometimes betrue. The possibility of gaming or manipulation of the information may alter the simple Jensen and Mecklingstory in interesting ways.When information may be manipulated, we might call it unreliable. This is a third category of attributes

of information that may have important implications for the allocation of decision-making. Another example isinformation that is noisy, or communicated with inaccuracy. Garbling (Geankoplis and Milgrom 1991) mightalso lead to decentralization of decision-making.A further notion is the degree of interdependence of the tasks needed to produce the output. Jensen andMeckling’s framework is useful for considering at what level in a hierarchical organization decision-making willbe located. But it ignores the fact that tasks performed throughout the production process may be highly interdependent(Thompson 1967; Lawler 1986). Production of different parts of a car may be modularized and insulatedfrom each other so that stoppages at one part of the process do not impact other parts. Decision makingover each component can then be decentralized to sub-units. A chemical production or oil refinery process, incontrast, has much more interdependent tasks, so there are large benefits from coordination. This leads into ournext topic, organizational structure and coordination.Organizational StructureBehavioral scholars often distinguish between “micro” and “macro” organizational structure. Micro involvesindividual job design, as discussed above. Macro encompasses the firm’s overall structure, including hierarchyand coordination mechanisms. (Decision-making falls in between, since it makes sense to think aboutpatterns of decision right allocation across the whole firm, but it is also an important part of individual job design.)Here we discuss macro organizational structure, long an important topic outside of economics, but notdiscussed much by economists.A key insight in this literature by Lawrence and Lorsch (1967) is that “one size does not fit all.” That is,there is not one ideal organizational structure, because improving the ability of a firm to perform along one dimensionwill diminish its ability to perform optimally along others. For example, producing at the lowest cost10for a given product design can impede the firm’s ability to respond quickly to changing demand conditions: theformer implies minimal R&D, while the latter implies much larger R&D. Thus, along the lines of our discussionsof external fit, a very interesting issue for economists to explore is how a firm’s optimal organizationalstructure varies with its environment, technology, etc.Much of the behavioral literature on structure can be crudely summarized as identifying conditions underwhich a traditional hierarchical functional organizational form is inappropriate (Galbraith 1977, 1995).14 Thetraditional organization puts a premium on command and control mechanisms designed to overcome principalagent problems. Workers are organized in departments according to function (manufacturing, sales, R&D, humanresources, accounting, etc.). This functional structure has several benefits: (a) it is clear which departmentis responsible for which tasks; (b) similar tasks are bundled into the same unit, facilitating specialization; (c)

there are clear career paths requiring investment in only specialized skills; (d) supervisors have knowledge ofand experience with subordinate tasks, aiding supervision and performance evaluation. The hierarchical functionalorganization is a natural extension of the traditional, Taylorist approach emphasizing specialization.There are important drawbacks to organizing along functional lines. A functional organization is bestsuited to firms that provide a single product or type of products that can be effectively managed centrally. Firmswith highly differentiated products (e.g., laptop v. mainframe computers; residential homeowner/car v. businessliability insurance) may find that creating separate divisions for each distinct category of products may be moreefficient, because it allows for further specialization within organizational units.15

Specialization in a functional organization comes at a cost similar to the costs from specialization in individualjob design: poorer integration, use of information, and learning across tasks. Consider a classic problem:product design, say in a computer company. Sales and marketing personnel know a great deal about whattradeoffs customers are willing to make among various features and cost; these employees have rich knowledgeabout customer demand curves. R&D employees have technical and scientific knowledge about what designsare feasible. Production employees have technical knowledge about the implications of various design decisionson costs. All of these kinds of knowledge are, to some extent, specific or costly to communicate to others. Butall are important sources of information for product design decisions. Decentralizing product design decisions toany of these three groups will result in non-optimal product design.Thus an interesting way to think about firm structure is to build on the idea of specific knowledge that iscostly to communicate. But instead of thinking about the degree of decentralization, the key issue is that thereare multiple “pockets” of specific knowledge located in various units in the firm. The firm would like to makecombined use of these pockets of knowledge effectively. There are two general aspects of the firm’s task (Lawrenceand Lorsch 1967). The first is to differentiate the knowledge into different departments within the firm, sothat departments specialize in developing and (to some extent using) the pockets of specific knowledge. Thesecond is to integrate the various pieces of knowledge so that they can be used together. This is done throughpolicies and processes that coordinate different units, allowing the pieces of knowledge to work together asneeded.16

Integration could be achieved by centralization, with the usual costs emphasized by Jensen and Meckling.But it might be interesting for economists to explore more fully what those costs are, and how they may bemitigated (especially in modern times with extensive use of information technology). Integration could also beachieved by decentralization, combined with coordination mechanisms designed to get disparate decision makers

to work together. There are two general approaches that firms use, incentives and coordination mechanisms,which we now discuss briefly.IncentivesThe classic economic solution to coordination is incentives, to motivate agents to work in concert witheach other. One approach is to give organizational units incentives to maximize overall divisional or firm value,though this tends to have free-rider problems. Another is to incorporate the effects of a manager’s actions onanother division into his or her performance measure. Still another is some form of transfer pricing. In fact,firms frequently use these incentive approaches.11One point here is that performance measurement and structure (breaking up into organizational units)are intimately related. Consider the case of a divisional manager given discretion over personnel policies for thedivision. The manager’s incentives will be strongly influenced by the divisional performance measure (e.g., costcenter, revenue center, profit center, etc.). This should flow through to how incentives are structured for the restof the division. Thus the divisional performance measure has important implications for incentives in the entiredivision. Moreover, the divisional performance measure the firm chooses will be closely related to how the divisionis defined; in fact, almost determined by how the unit is defined. Definition of the unit defines how performancedata (costs, revenue, headcount, etc.) are aggregated and computed. Therefore, the division of the firminto sub-units immediately gives rise to certain performance measures and incentive effects.17

Unfortunately, it seems unlikely that the incentive approach to coordination will be completely effectivein many situations, by the very nature of the problem. If knowledge is specific, or costly to communicate, it ishard to imagine how to develop performance measures (at least, ex ante and reasonably objective measures) thatmotivate agents who do not possess this knowledge to act in ways that take it into account (also see footnote 16above). Therefore, other approaches are also likely to be important.Coordination MechanismsAn alternative for achieving coordination across pockets of specific knowledge is to use one of a plethoraof structures that overlap traditional functional hierarchy. These structures may be permanent or temporary.One example, often used in product design, is a cross-functional team, with members from various units thatpossess information that can be combined productively. Another is a matrix organization, in which employeeshave two hierarchical assignments, one functional and one by product, project, etc. Galbraith (1995) describes acontinuum of lateral coordinating mechanisms:Informal (or voluntary) processes occur spontaneously; they are not formally established bymanagement;Teams are formal groups that are used to complement the informal voluntary processes; they areestablished by management;Integrators are full-time leaders who are appointed to lead the formal groups, and are equivalent

to cross-functional managers responsible for managing a product, project, brand, etc.Galbraith describes “informal” organizations as those “characterized by voluntary coordination acrossunits” (p. 49). He suggests that informal processes are usually preferable, because formal structures have to bedesigned and managed, with attendant principal agent problems. However, this does not seem so clear to aneconomist. Informal mechanisms may clearly work against the firm’s interests if the incentives of the initiatorsare not well aligned with firm objectives. In this sense Galbraith’s definition of an informal organization assumesthat the incentives induce only those voluntary actions that are aligned with firm objectives. A contrastingview (Lawler 2000), and one to which most economists are sympathetic, is that reward systems can create lateralinterdependence by giving everyone the same fate, i.e., by explicitly aligning their financial interests witheach other.According to Galbraith, an organization can foster voluntary coordination by: interdepartmental rotation;interdepartmental events; co-location; information technology; mirror image departments, and consistentreward and measurement practices. Is there a meaningful distinction between “voluntary” and “formal” coordination?If the only distinction is the definition within the organizational chart, then the distinct may be irrelevant.But if there is really something more, then economic tools like agency theory could help better define thedistinction between informal groups formed voluntarily (via consistent reward systems) and formal teams set upby management using explicit team rewards.One insight from behavioral empirical research on coordination mechanisms (e.g., teams) is the relationshipbetween the frequency with which a team meets, and its design. If team members meet only infrequently,they can be a cross-functional team imbedded in a firm organized along traditional, functional lines. Ifthey have to meet frequently, then it may be better to have team members’ primary reporting relationship be with12the team itself, and the secondary reporting relationship be with functional managers who help monitor the rolesthat specialists perform in teams. Specialization is preserved to some extent, but the efficiency of informationflow between the different areas of specialization within the organization is improved and, consequently, thespeed of responsiveness to the market. Thus, the extent to which a mechanism is permanent and emphasized,relative to the underlying functional structure, seems to depend on the relative importance to the firm of integrationof the pockets of specific knowledge.18

In any case, this points to the basic tradeoff in using coordination mechanisms: they reduce the employee’sspecialization and introduce monitoring and performance measurement costs, but they broaden the employee’sview of tasks and facilitate learning and collaboration. In other words, optimal organizational structure,

with functional hierarchy and coordination mechanisms, is based on much the same issues as is job specializationor enrichment.TeamsTeams are a topic that has received considerable attention outside, but little inside, economics. Examplesof different types include work, parallel, project and management teams (Bailey and Cohen 1997). A large partof the literature is empirical, focusing on how teams can function effectively. While the topics of this empiricalliterature are interesting (e.g., the role of interpersonal skills as a type of human capital), it is the theoretical roleof teams that is relevant for the present discussion.19

Early work on teams was motivated by the socio-technical literature, which examined how technologyeffects on work organization are mitigated by the social structure and needs of the workers (Trist and Bamforth1951; Miller and Rice 1967). This approach led to designing work to be performed by groups, to take advantageof the benefits of social interaction (Hackman and Oldham 1980).Economists tend to view teams as a set of workers with a common task or performance measure, but saylittle beyond such stylized representations. Clearly there is more to teams than that. For example, empirical evidencesuggests that teams are more likely to be used when firms adopt job enrichment (MacDuffie 1995). Ourdiscussion of job design suggests a possible explanation: enriched jobs are designed to facilitate learning andcontinuous improvement by workers. Ideally workers need to know all aspects of closely related tasks for thispurpose (that is, the job should be broadened up to the level of modularization of production). But in many casesthat would imply job enrichment so broad that specialization and productivity would suffer. A balance betweenlearning through job enrichment and specialization can be struck by using teams that work together.There are several ways in which teams can facilitate learning. One is cross training. Another is informationsharing: one worker may have knowledge that can improve the productivity or decisions of another. A thirdis collaboration, in coming up with initiatives, testing them, and making decisions. An interesting avenue formodeling teams would be to explore the dynamics and incentives of collaboration in a group. The economictools of information economics, agency theory, and game theory seem particularly well suited to the task.According to Lawler (1986), the emphasis on teams versus individual job enrichment depends on boththe technology and task uncertainty involved in producing the firm’s output, and on the individual’s needs forlearning and social interaction (H&O’s “growth need strength” moderator); see Table 1. Thus, the firm needs toconsider external fit, and there may also be internal complementarities between job design and personnel policiessuch as recruiting and training. The table also suggests how characteristics of workers (bottom panel) interactwith strategic objectives (top panel). If the firm wants to produce a product that requires a great deal of interdependent

work, an assembly line approach with narrowly defined jobs will not work. Two alternatives arehaving the product produced by teams or by individuals who do an entire “assembly” by themselves. Which isbest depends on the technology, and the firm’s ability to either mold the organization to the characteristics of theincumbent workforce, or to change those characteristics through a training, restructuring and turnover.13Organizational DynamicsAs noted in the introduction, the systems view of organizational design emphasizes that organizationsare dynamic, evolving over time as their environments change. We now briefly discuss ways economists canthink about organizational dynamics and design.Organizational EvolutionIn Figure 1, there is a feedback from Business Outcomes to Strategy. This captures the idea that thefirm’s existing design may limit its optimal strategy and ability to evolve. This is a strong theme in organizationalresearch outside economics, where “Managing Change” is considered a fundamental topic. This literatureis very large; here we focus on a few basic issues likely to be of greatest interest to economists.Consider again our hypothetical firm with an existing strategy and design. Suppose the environmentchanges in some important way, such as a new technological development. If we were to consider the firm’s optimalorganizational design ignoring its existing organization, the problem would be unconstrained. But an interestingquestion is the degree to which the existing organization constrains the firm’s ability to change orevolve. It is quite likely that this is the case, for several reasons.The first constraint is that the firm has made organizational investments of various kinds, some of whichare costly to change or abandon. For example, the firm has made investments in matching and developing appropriateemployee skills. If there are turnover costs (say, because of the legal environment, severance provisions,or search costs), then changing to a new workforce with new skills may be prohibitively costly. Similarly,many firms invest in a reputation for how they treat employees. This facilitates implicit contracting over employment.Any organizational change may undermine this reputation, giving firms an optimal reluctance tochanging policies.A second possible constraint on organizational change involves complementarity, or internal fit betweenpersonnel policies. If the firm has a bundle of policies that have been designed or evolved to work well together,and it wants to change one of them, it may find that they are no longer complementary. If so, the firm wouldface three options: (1) not changing policies, reducing external fit of the organizational design with the strategy;(2) changing the strategy to better match existing polices and practices; and (3) changing to a new bundle ofcomplementary policies.The problem with (1) is that it is likely to reduce the firm’s competitiveness. The problem with (3) isthat it may be costly to change multiple policies, structure and processes simultaneously (or nearly simultaneously).In some cases, the firm may instead choose (2). In fact, this is essentially the core competence idea in the

strategy literature (Prahalad and Hamel 1990): a firm should craft its strategy around what the organization isalready good at. This is an intriguing argument that is worthy of careful theoretical (and especially empirical)research.20

A possibly even more intractable setting for organizational change involves two firms that decide tomerge for strategic reasons. Both firms have existing job designs, structures, coordination mechanisms, decision-making, employee skills, and implicit contracts. How are these two organizations to merge? If it is a mergerof similar firms within an industry (i.e., a consolidation merger), this necessitates not just changing one organization,but two, and worse, blending them together. Blending is also necessary when firms in different industriesmerge. But the degree of blending in this case may be mitigated by the need to preserve unique aspects of thefirms’ original organizations in order to realize synergies. The costs of doing so effectively suggest that, in manymergers, one firm’s organization will be left relatively intact, the other’s largely abandoned (including its employees),and the customer lists merged. Casual empiricism suggests that this is often the case. This would befeasible for mergers that involve marketing synergies, but possibly not for other kinds of mergers. This issuepotentially could be examined empirically by collecting personnel records for merging firms before and after themerger, to see which groups of employees survive.14Implicit ContractingEconomists have long recognized that implicit contracts play an important role in the labor market. Thestandard economic approach focuses on the limitations of formal contracts and what that means for how wagesare set and how firms elicit optimal effort and investment from workers. Classic examples include risk sharingbetween the firm and workers over macroeconomic fluctuations, and sharing investments in firm-specific humancapital. Another is the possible use of a wage premium to elicit effort that the employee might otherwise be reluctantto put forth (Akerlof and Yellen 1986).Despite the rich variety of applications for which economists have used the notion of implicit contracts,they typically ignore organizational dynamics. For example, a key element of the “effort” that firms want toelicit from employees includes suggesting process improvements, which in some cases might eliminate theworkers’ jobs. This is a significant omission, given the importance of such feedback in optimizing productmanufacturing and delivery mechanisms. Evidence from the behavioral literature on high performance worksystems shows that firms sometimes use job security provisions to elicit the desired information. Job securitycan never be an ironclad explicit guarantee, so implicit contracts have a role to play.Behavioral researchers have analyzed implicit contracts with renewed interest in recent years, under the

heading of “psychological contracts” (Rousseau 1995). Two theoretical issues are particularly relevant. The firstis that differences in the degree of uncertainty in the environment faced by organizations lead to different implicitcontracts (the idea of external fit once again). A startup company (or division of a larger company) withuncertain market prospects cannot offer the same implicit long-term employment as a company with a history ofstable or growing demand for its product. A small company cannot offer the same career/skill development orpromotion opportunities as a large company.The second interesting theoretical issue is an assertion (Rousseau 1995) that increases in the degree ofuncertainty in the environment (say, due to increased competition or technological change) are likely to lead toincreased reliance on “transactional” implicit contracts. Such implicit contracts emphasize the immediate benefitsof exchange between the worker and firm. This is in contrast to the deferred benefits available from “relational”implicit contracts, which extend over longer periods (see below). For example, industries that previouslywere highly regulated and/or protected typically cannot offer the same level of job security after deregulation orthe lifting of protection.Working against this logic, however, is that an environment of increased uncertainty may make implicitoffers by the firm less credible. Economists (usually game theorists) analyzing implicit contracting and corporateculture often emphasize the benefits of stability, consistent, long history, and related concepts in strengtheningthe reputation of the firm to employees (Kreps 1990; Camerer and Vepsalainen 1988). We can think aboutthe employee’s gauging of employer credibility over implicit promises as a statistical inference problem. Stabilityand consistency of how the firm has treated employees mean lower variance, or greater precision, of employeepredictions about future treatment. Similarly, longer history implies more data, and therefore more accuratepredictions. When the environment changes suddenly, the firm may lose much of its reputational capital.Rousseau (1995) uses two dimensions to frame the discussion of implicit contracts: expected job durationand whether there are explicitly specified performance terms for the contract. This leads to four types ofcontracts:Relational: long employment duration, performance terms not specified explicitly. Examples includetechnical and professional employees, and managers;Balanced: long duration, performance terms specified. Examples include “core” blue collar andoffice support occupations;Transactional: short duration, performance terms specified. Examples include temporary employees,contract employees, and consultants;Transitional: short duration, performance terms not specified. Examples include transactionalemployees who are auditioning for long-term employment (e.g., temp-to-perm), and employees15

of an organization (both relational and balanced) undergoing restructuring and/or implementingsignificant headcount reductions.It has been widely noted that increased uncertainty in the environment has led many companies to decreasefixed labor costs by converting relational and balanced jobs into transactional and transitional jobs. Anecdotalevidence includes the end of “lifetime employment” at traditionally stable companies (such as IBM andGeneral Motors), the removal of entire layers of middle management, and increased use of temporary and contractlabor for jobs that formerly were staffed by “core” employees.An interesting issue is whether such changes impact a firm’s ability to deliver the product dictated by itsstrategy, because of problems with external fit. If the majority of the firm’s employees were hired under a regimeof implicit contracts that promised long-term job security, those same workers might not be happy with theconversion to transactional or transitional status. At the extreme, if the resistance to implicit contract changes byincumbent workers is too strong, a firm might not be able to make the transition to a new technology or processfor organizing work without suffering a significant drop in productivity. The alternative is to open an entirelynew site and staff it with new hires and incumbent workers who opt for the new implicit contract. This is consistentwith Ichniowski and Shaw’s (1995) evidence on the greater use of the complete set of changes that make uphigh performance work systems by new sites; older sites, in contrast, are more likely to adopt only a subset ofthe changes.Knowledge ManagementAt the intersection of implicit contracts and organizational evolution is the issue of knowledge management.In intellectual work, employees develop specific knowledge that is not entirely firm-specific, but is a formof intellectual capital or trade secrets that are valuable to the firm.21 For example, a software engineer may developtechniques that would be beneficial to the firm’s competitors. The problem arises because of possible employeeturnover. First, can the firm create effective incentives to encourage the transmission of the specificknowledge by the employee before he or she leaves? Second, can the firm protect its investments in intellectualcapital from expropriation by the departing employee? Here we focus on the second issue, since personnel economicshas a long history of thinking about ways to reduce turnover (see Lazear 1998).Knowledge management and retention of key employees have received a lot of attention within themanagement literature (and among consultants) for two reasons. The first is the current relatively high demand(and low unemployment rate) for high skilled technical and professional employees. This has greatly increasedlabor market opportunities for these workers, leading to increased turnover. The usual way to reduce turnover isto offer deferred compensation that is not vested, but this is harder for firms to do successfully if the employee’s

outside alternatives rise rapidly. This is bound to be a transitory phenomenon, because markets will adjust eventually.22

The more relevant issue is the long-run trend toward more knowledge-intensive work. How can a firmprotect its intellectual capital? One possibility is to recognize the difficulty of preventing the employee fromleaving and taking the knowledge. In this view, the intellectual capital is simply a form of general human capital.But the issue becomes more difficult if the employee’s knowledge is in the form of trade secrets, as seemslikely in the case of R&D employees. The optimal legal environment, restricting the employee’s post-firm employmentrights, is far from obvious. The issues are somewhat similar to those of patents in general. But generalreluctance of employment law to restrict individual employee rights makes it more difficult for firms to protecttheir part of investments in such intellectual capital.Designing in AdaptabilityThe discussion of organizational evolution hinged on the case where the firm’s environment changesunexpectedly. But there is another sense in which a firm can think about evolution. It can design personnel policieswith adaptability in mind, recognizing that the environment is likely to change (in some ways that are partiallyforeseeable, like future technology changes, and in some ways that are not foreseeable). This focuses on16Intermediate Outcomes, or attributes of personnel policies, that economists do not usually consider. Here webriefly mention some attributes that are likely to facilitate organizational adaptability.First, our discussion of job design above emphasized the development and use of knowledge throughjob enrichment and related practices. Thus, job enrichment and, in some cases, decentralization of decisionrights to employees, are ways that firms can develop some adaptability. Job enrichment facilitates continuousimprovement by employees, which is more beneficial in settings where the firm’s product, product line, or productiontechnology change more rapidly.“Empowerment” of certain decision rights allows the firm to tap into suggestions that may flow out ofnew information that comes to employees. This builds in a way for the firm to react to and exploit useful specificknowledge that arises in various pockets throughout the organization. This is especially true when whatFama and Jensen term (1983) “decision management” rights are decentralized. These are the rights to suggestnew initiatives, and to decide how to implement new initiatives that have been chosen. Thus, we predict thatdecision management is more likely to be decentralized, the greater the change in the industry (in the sense ofunpredictability, technology, competition, regulation, etc.).23

Another approach to adaptability is to invest in more adaptable employees (Lindbeck and Snower2000). It is usually argued that giving workers a broader set of skills (either more skills, or including intellectualas well as physical skills) allows them to be redeployed more readily.Implicit contracts with employees are clearly very important to adaptability. The firm would like to motivate

employees to suggest new ideas, reveal specific knowledge, and be willing to be redeployed as circumstanceschange. To some extent these can be motivated by incentives and other practices, but they are likely tobe affected to a great extent by implicit contracts. Thus, a firm culture that encourages suggestions, collaboration,some risk taking, and flexibility by employees would seem desirable in most cases.24 Countering this is theproblem of how to encourage employees’ willingness to share knowledge in those cases where the firm is notable to offer plausible long-run employment promises.Methodological and Empirical IssuesComplementarities?Personnel economists have recently become very interested in internal fit, or complementarities amongorganizational policies. Much of this interest has been theoretical, stemming from Milgrom and Roberts’s (1990,1995) application of lattice theory. There has also been a strong interest in empirical work on internal fit in recentyears, both inside and outside personnel economics (Ichniowski, Shaw and Prennushi 1997; Levine 1995;MacDuffie 1995; Cappelli and Neumark 1999). Of course, at some level, it is inarguable that a firm should designpersonnel policies taking other policies into account, ceteris paribus. The real question is whether complementarity(or substitutability) between two or more specific policies has important effects on firm objectives likeproductivity.We think that economists’ interest in complementarities is warranted. If the view that internal fit amongpolicies can be an important source of organizational efficiency is correct, then this is important for us to understandand teach. Moreover, behavioral scholars have argued for decades that this part of the systems view is important.Much of their argument is based on case studies and their experiences working with real organizations.This is not the kind of source for empirical insights that economists are used to using. However, there are manyinstances where economists have learned of important empirical insights or topics from other fields in this way(in the spirit of the first quote at the beginning of this chapter). Given the strength of this thread through the behavioralliterature, economists should take this idea seriously.Some surveys have argued that there is now strong evidence for the impact of internal fit on organizationaleffectiveness. While we believe that existing evidence is provocative and promising, we remain somewhatmore skeptical, and argue that we have some way to go before we can be confident about the empirical validityof this concept. Consider, for example, the case of Lincoln Electric Co. (Berg and Fast 1975), the best-selling17Harvard Business School case of all time. Economists often describe Lincoln Electric as an example of the valueof internal fit. Lincoln makes extensive use of piece rate incentives. They also have other policies designed to

deal with classic problems with piece rates, such as poor quality, lack of cooperation, etc. In particular, Lincoln’ssuccess is often attributed to its ability to use complementary implicit contracting practices that ensure effectivesubjective performance evaluations (Baker, Gibbons and Murphy 1994). These practices are said to be especiallyeffective because of Lincoln’s long and consistent history. Indeed, Lincoln has experienced strong economicsuccess, and few of the classic incentive system problems, for many years.25

However, consider another case study on piece rate incentives, Safelite (Lazear, 2000). Safelite institutedpiece rate incentives for automobile windshield installers. Lazear’s analysis indicates that productivity skyrocketed20-40%, depending on how the effects of worker selection are estimated. These numbers are similar tothe productivity differentials attributed to Lincoln’s set of personnel policies. Yet Safelite did not obviouslyinstitute a system of complementary policies of the form seen at Lincoln. Nor did the company have strong implicitcontracting and subjective performance evaluation, or a long history of consistent treatment of workers onwhich to base implicit contracts.The disparity between these two case studies immediately makes one wonder how important complementaritiesreally are to the effectiveness of personnel policies. Of course, these are only two observations, andthe Safelite case was not written with a broad description of other policies in mind. Much preferred would besystematic empirical research with larger datasets, such as the works cited above.One problem with empirical research on complementarities stems from the lack of strong theoreticalanalysis of the issues. Unless we can develop a convincing ex ante theory of precisely what policies are supposedto be complementary with each other, empirical work has little guidance for what set of policies to lookfor. Suppose that, following the literature on strategic human resource management, we observe that firms usinga certain set of policies together tend to have higher productivity, lower costs, lower turnover, etc.26

Are we toconclude that these policies exhibit complementarity or internal fit? Another possibility is simply that the policiesare correlated with something else. Perhaps the policies just are often intelligent ones to use, so their use inpractice is driven by correlation with managerial talent. Another possibility is that they are driven by some environmentalvariable. Perhaps a number of policies are all well adapted to certain production environments, butare not complementary with each other per se. Then a firm that uses this bundle of policies will have higher productivity,but not because of internal fit.Therefore, before further empirical work on complementarities is done, we urge economists to thinkmore carefully about the theory behind it. We need to develop more rigorous theories of exactly which policiesare complementary with each other, and why. We also need to think more carefully about how to disentanglecomplementarities empirically from other effects that might drive mutual adoption of a set of policies (though

consistency of empirical patterns with a well-respected theory would be a substantial step in that direction).One way for theorists to tackle this problem is to consider Baron and Kreps’s (1999) assertion that internalfit can lead to patterns of organizational design. If we can develop a theory that suggests that policies A, Band C should be observed together, while policies D, E and F should be observed together, but that firms shouldrarely use A, B, or C with D, E, or F, this would be an interesting and new test.27 But that still begs the questionof what determines whether a firm adopts one pattern or another. An even stronger theory would yield predictionson this dimension, so that we could test whether the incidence of any observed patterns (if there are suchorganizational design patterns) varies as predicted.Thus, we also need to think about factors that might drive patterns of policies. We think the answer hereis to consider external fit. For example, our discussion of job design suggested that there might be two roughpatterns, enrichment (with decentralization of some decision making) or specialization. It also suggested that theenrichment approach is more valuable in information rich environments, where there is large scope for on-thejoblearning. We described several examples of when a firm might have that kind of environment; e.g., when afirm has made less use of industrial engineering; if the product line is relatively new; the product is complex; ortechnology is rapidly changing. These are testable predictions about the relationship between the firm’s environmentand it’s the adoption or effectiveness of a pattern of personnel policies.18Empirical ResearchPersonnel economics is a productive and empirically relevant field. Great strides have been made, particularlyin developing a theoretical structure that provides a rich framework for analyzing organizational issues.However, empirical work in the field is sparse and too narrowly focused. There are a large number of empiricalstudies of careers and compensation, but limited work on most other topics. For personnel economics to continueto grow and be influential, it is paramount that researchers focus more on empirical research.There has been growing interest in collecting new types of data, especially personnel databases (e.g.,Medoff and Abraham 1980; Lazear 1992; Baker, Gibbs and Holmstrom 1994a,b; Lazear, 2000). Such databasesare rich sources of information on the internal workings of firms. However, they have two limitations. First,these datasets are usually collected from a single or small number of firms, so generality of findings is an issue(though matched worker-firm datasets are starting to be collected, especially in Europe; see Abowd and Kramarz1999). Second, by their nature personnel databases focus on some variables (those collected for personnel computersystems) but exclude others.The most common approach to solving the generality problem is to collect personnel data across a large

sample of firms, typically in the same industry (Ichniowski, Shaw and Prennushi 1995). This substantially reducesthe problem, though it does not entirely eliminate it if the data are from only a single industry. Unfortunately,such data collection is difficult and expensive, since it involves contacting and collecting data from manyfirms at once. Nevertheless, we hope to see more projects along these lines in the future.A second problem with existing empirical work by economists is that it tends to ignore a wide variety ofimportant and interesting issues, such as those discussed above. Some of the most interesting empirical workwill require collecting new kinds of variables. Many of these variables are difficult to measure by traditionalmeans. Others are subjective by nature. Nevertheless, it is important for economists to find ways to collect andanalyze such data systematically, instead of leaving this area solely to behavioral researchers.The most likely way to collect such datasets would be to write and administer surveys. Survey researchis standard in most fields outside economics.28 By using surveys, economists can begin to do empirical work ondecision-making, job design, organizational structure, implicit contracting, etc. For example, Cooper (1998)used survey techniques to measure employee cooperation, linking that to compensation contracts and other variables.Behavioral researchers have developed a variety of methods to measure subjective dimensions of employment.For example, job satisfaction and commitment measure worker perceptions of the employment relationship;organizational citizenship behaviors (OCBs) measure the ways that they express their perceptionsthrough taking actions that help the organization even if there is no immediate monetary reward (Mobley 1982;Hom, et al, 1992). It would be interesting to interpret such data from the perspective of economic theories ofimplicit contracting, corporate culture, and repeated games. More generally, an extensive body of work by behavioralresearchers links motivation and both effort and job performance (e.g., Whyte, et al. 1955; Steers andPorter, 1991). It might be interesting to relate such variables to how incentive contracts are designed, along thelines of typical economic studies that link incentives to quantitative output and cost measures.ConclusionEconomics has added much to our understanding of personnel management and organizational design,as have behavioral approaches. Economics uses formal mathematical modeling more than other fields. Suchmodeling has the benefit of providing a systematic way to derive conclusions about topics that otherwise mightbe too difficult to analyze formally. In the process, though, economics sometimes ends up throwing out muchthat is interesting. It does this in two ways. First, economic theory may be applied in areas where economistshave already developed extensive insights. Thus, we have a vast literature on incentive theory. Second, empiricalwork in economics tends to follow traditional econometric methods, and focus on traditional econometric databases.Unfortunately these rarely have much interesting within-firm information.19

Behavioral researchers tend to take a systems view of organizational design, which is one reason whythey approach the field differently than do economists. In recent years economists have shown increasing interestin part of this view, internal fit (which they term complementarity), in recent years. This is an important conceptthat deserves careful scrutiny. However, economic work on the idea of complementary has far to go. Empiricalwork is suggestive, but hardly conclusive. The theoretical underpinnings are scant. This issue wouldbenefit from more thorough mathematical modeling, and more data collection.In order to make progress, a logical starting place would be to bring in the related concept of externalfit. It is sometimes argued that complementarities imply patterns of organizational design. Presumably, eachpattern would be more useful for some purposes than for others. Considering how a firm’s policies help achievevarious intermediate outcomes, such as learning, adaptation, cost reduction, etc., might be an important way tobegin modeling internal fit as well.Economic theory of the internal design of firms would benefit from a richer view of what actually happensinside firms. Economic theory is highly abstract about jobs and other features of organizations. Of course,this has great value in developing generalizable insights. However, it means that economists have little to sayabout a great many important and practical topics. Economics could learn from other fields by paying attentionto what behavioralists have found to be interesting topics.We described several areas of organizational research where there is (yet) little economic work. Theseinclude intrinsic motivation, job design, decision-making, coordination, and more specificity about implicit contracting.Notable about this list is how fundamental these issues are to organizational design. Moreover, most ofthem have to do with information, communication, and incentives. Economists should have much to say aboutthese topics; we sketched only a few starting ideas.Our discussion of how to conceptualize the firm had a theme: it is useful to view the firm as a developerand processor of knowledge. Many economic models use a physical task metaphor (such as the emphasis oneliciting “effort” in incentive models). Yet much of what is interesting about organizations is intellectual. Iffirms can effectively invest in industrial engineering, perfecting the process in advance, and if the product andmarket are stable, then it is reasonable for the firm to use a Tayloristic approach. The firm should then lookmuch like Adam Smith’s pin factory. But that does not apply to the majority of firms. There are fascinating andfundamental issues involved in collecting, processing, learning from, and communicating information within thefirm. These issues are also important in understanding how organizations adapt and evolve.The most interesting progress in personnel economics is likely to occur in empirical research. There aremany important and interesting topics about which we still know little to nothing empirically. And, as Lazear(1999) argues, personnel economics must be empirical to be relevant in the long run. We need to collect more

data. We need to collect more interesting data. In order to do so, we also will sometimes need to employ lesstraditional data collection methods, such as surveys. These are not always easy or low cost tasks, but the returnscan be enormous.Footnotes1. We are thinking about several overlapping fields. Lazear’s (1999) definition of personnel economics is best for our purposes: “The useof economics to understand the internal workings of the firm.”2. Behavioralists distinguish between “policies” and “practices.” The former indicates norms that are formally codified as officially sanctioned,while the latter indicate norms that may or may not be formally codified. Economists do not make such distinctions. We use eitherterm to refer informally to both.3. A notable exception is the long line of research in labor economics that tries to explain why workers with easily identifiable characteristics(union status, exempt status, race, ethnicity, gender) have different observed outcomes (wages, turnover, training). The motivationfor this lies primarily with governmental policies on collective bargaining, overtime, benefits, and anti-discrimination laws. However, thisobjective has led economists to use nationally representative data sets to the virtual exclusion of within-firm analysis. Of course, there areexceptions, such as Doeringer and Piore (1971), and Levine (1995).4. Of course “policies” themselves may be explicit or implicit, and intentional, historical, ad hoc, or accidental.5. The “pins” were actually nails. Interestingly, Smith’s analysis of specialization may trace back to Persia 900 years ago (Hosseini 1998).6. As we were completing this draft, Lindbeck and Snower (2000) published a very interesting analysis of multi-task job design that isclosely related to our arguments. Their analysis provides a modeling framework for thinking about these issues, and shows how changesin four factors can lead to greater use of job enrichment: changes in production and information technologies that promote task complementarities,changes in worker preferences, and advances in human capital that make workers more versatile. The first two are mostclosely related to our arguments. Our analysis is complementary to theirs in several ways. They present a stylized, technological view ofmultitask production, while we provide some theoretical underpinnings about intrinsic motivation, learning on the job, and what mightspecifically drive task complementarities. Our emphasis on learning and complexity of the work environment also more fully fleshes outan explanation for why there has been a recent trend toward job enrichment.7. H&O have a related idea in their model, “Task Significance,” which they do not emphasize. This can mean several things. They describeit as “the degree to which the job has a substantial impact on the lives of other people”; obviously this might affect the worker’sutility function through altruism. Task Significance could also mean the degree to which the job affects utility directly (Murdock 1998).An economist might add that a task could be significant to the worker if output affects the worker’s pay, through incentive compensation.8. It is worth noting that there is a debate over the causal factors behind intrinsic motivation; see Ryan and Deci (2000) for a (brief) review.According to Ryan and Deci, self-determination theory holds that intrinsic motivation is an innate propensity of the individual thatcan only be enhanced or diminished by external factors. According to this view, “all expected tangible rewards made contingent on taskperformance … undermine intrinsic motivation” (p. 70). If indeed this were the case, then job enrichment undertaken to enhance intrinsicmotivation would be doomed to failure.9. In the Tayloristic approach, workers are a source of variation and error; thus they lead to quality problems and cost increases. In itsideal form, industrial engineering squeezes humans out of the process and uses mechanization instead. This is in striking contrast to themore modern view: workers can be a source of ideas for improvement, quality, cost reduction, etc.10. Note that it has nothing to do with intrinsic motivation and psychology. Further, the kind of intrinsic motivation that H&O argue isincreased by job enrichment (mental involvement in the work) is ideal for decentralization and learning by workers. Thus, the H&O

model overall has less to do with psychology than might initially seem apparent.11. For an interesting economic view of Total Quality Management programs (which usually employ “quality circles” or other forms ofteams) as attempts to have workers engage in the scie ntific method to develop process improvements, see Jensen and Wruck (1994). Inour view TQM is a special case of more general job enrichment methods designed to get employees to generate and use knowledge toimprove quality and efficiency.12. See Garicano (2000) for an interesting approach emphasizing costs of processing and transmitting knowledge.13. Fama and Jensen (1983) provide an interesting extension, distinguishing between different types of decisions. They describe a fourstageprocess: generating ideas or initiatives; choosing/ratifying initiatives; implementation of choices; and monitoring of implementation.They term the first and third decision management, and the second and fourth decision control. The distinction gives a richer way ofthinking about decision-making and decentralization, but space limits prevent us from exploring it here. As a quick point, note that firmstend to use relative centralization of decision control, to improve coordination, combined with relative decentralization of decision management,to improve use of specific knowledge.14. Galbraith (1995) provides a light treatment targeted toward practitioners, sufficient for an overview. Galbraith (1977) is a more academictreatment.15. Firms often organize along lines other than functional, e.g., product, geographical, or customer divisions. However, such structuresalmost always have a strong element of functional hierarchy within them. Breaking up into sub-hierarchies in this sense appears to be dueto limitations on the optimal size of a hierarchy, such as communication costs, garbling of information, limits of managerial talent, etc.16. The need to integrate economic activity across several decentralized agents with specific knowledge may be one reason for the existenceof firms instead of markets. Price mechanisms may be inadequate for coordination when pockets of knowledge are costly to communicatebut need to be integrated. This same logic suggests that incentives (internal substitutes for the price system) are unlikely to fullyachieve coordination inside a firm. If they could, it would beg the question of why the activity is internalized instead of using markets.2117. See Lawler (2000) for a discussion of the tradeoffs involved.18. At the extreme, if such cross-functional teams perform a majority of the critical work of the organization, then they may become theprincipal unit that defines the organizational structure (Mohrman, et al, 1995).19. See Bailey and Cohen (1997) for a detailed review of the empirical literature.20. A flip-side to this argument is that if organizational investments are costly or time consuming, they might be a source of competitiveadvantage (a barrier to entry) if they are well matched to the firm’s environment. For example, firms with long histories in relatively stableindustries should have competitive advantages against new entrants, ceteris paribus.21. Thus, the intellectual capital is firm-specific in the sense that it involves specific knowledge of the firm’s technology, organization,methods, or products. But it is general in the labor market sense that it increases the employee’s outside market value, because competitorscan benefit from learning this knowledge. It is different from the usual general human capital case, because the benefit to the departingemployee or competitor comes at the expense of the firm. Thus, there may be an efficiency loss when the employee departs, similar tofirm-specific human capital. In this way it is similar to the transferable skills described by Stevens (1996).22. David Finegold points out that, in the specific case of the demand for technical workers in the U.S. (especially software programmers),rapid and substantial rises in hourly wage rates during the 1990s did not lead to an adequate supply response. Lags and rigidities inthe education system and immigration policy are the most likely reasons.23. The method of evaluating initiatives also has implications for the firm’s adaptability, since it affects the number of projects that areevaluated, the percentage that are accepted or rejected, the likelihood of Type 1 and Type 2 errors, and the overall “creativity” of firmprojects (Sah and Stiglitz 1986).

24. Not all of these elements are desirable in all circumstances. For example, encouraging suggestions on process improvements from thecleaning staff at a factory producing explosive chemicals can be beneficial; but encouraging risk taking among these same staff could bequite detrimental. Similarly, changes to the production process at such a factory are only valuable if they can be codified and replicated(through the identification of best practices). So it may be optimal for management to limit the number and duration of times duringwhich suggested deviations from standard operating procedures are vetted.25. Though the Lincoln Electric case is now 25 years old, it remains relevant to the company’s U.S. operations today. The policies inplace in the U.S. as late as 1998 were almost identical to those described in the case. Though Lincoln had substantial difficulty in itsoverseas operations in the last decade or so, for a variety of reasons, its U.S. operations continue to operate with very high productivity,etc.26. Huselid (1995); MacDuffie (1995); Lawler, et al, (1998); Guthrie (forthcoming).27. An additional test would be to look for cases of organizational change, and see if firms adopt theoretically complementary changessimultaneously, or if doing so increases effectiveness of the change. This approach has yielded mixed support for the theory of internal fitso far (Ichniowski and Shaw 1995; Lazear 2000).28. In fact, most large datasets used in labor economics are also collected by government surveys, but economists have been insulatedfrom the data collection process.ReferencesNote: additional references are in Appendix A.Abowd, John and Francis Kramarz. 1999. “The Analysis of Labor Markets Using Matched Employer-EmployeeData.” In Handbook of Labor Economics, vol. 3B,eds. Orley Ashenfelter and David Card, Chapter 40, pp. 2629-2710. Amsterdam: North Holland.Akerlof, George and Janet Yellen. 1986. Efficiency Wage Models of the Labor Market. Cambridge: CambridgeUniversity Press.Bailey, Diane E. and Susan Cohen. 1997. “What Makes Teams Work: Group Effectiveness Research from theShop Floor to the Executive Suite,” Journal of Management 23(3): 239-290.Baker, George, Robert Gibbons and Kevin J. Murphy. 1994. “Subjective Performance Measures in Optimal IncentiveContracts.” Quarterly Journal of Economics 109: 1125-1156.Baker, George, Michael Gibbs and Bengt Holmstrom. 1994a. “The Internal Economics of the Firm: Evidencefrom Personnel Data.” Quarterly Journal of Economics 109(4): 881-919.Baker, George, Michael Gibbs and Bengt Holmstrom. 1994b. “The Wage Policy of a Firm.” Quarterly Journalof Economics. 109(4): 921-955.Beckmann, Martin. 1977. “Management Production Functions and the Theory of the Firm.” Journal of EconomicTheory 14(1): 1-18.Becker, Gary S. and Kevin M. Murphy. 1992. “The Division of Labor, Coordination Costs, and Knowledge.”Quarterly Journal of Economics 107(4): 1137-1160.Beer, Michael, Bert Spector, Paul Lawrence, D. Quinn Mills and Richard Walton. 1984. Managing Human Assets.New York: Free Press.Berg, Norman and Normal Fast. 1975. The Lincoln Electric Company, case #376-028. Boston: HarvardBusiness School Press.

Calvo, Guillermo and Stanislaw Wellisz. 1979. “Hierarchy, Ability, and Income Distribution.” Journal of PoliticalEconomy 87(5): 991-1010.Camerer, Colin and Ari Vepsalainen. 1988. “The Economic Efficiency of Corporate Culture,” Strategic ManagementJournal 9: 115-126.Cappelli, Peter and David Neumark. 1999. “Do ‘High Performance’ Work Practices Improve Establishment-Level Outcomes?” NBER working paper 7374.Cooper, Joseph. 1998. “The Effect of ESOPs on Effort and Cooperation.” Working paper, Tuck School, Dartmouth.Doeringer, Peter B. and Michael J. Piore. 1971. Internal Labor Markets and Manpower Analysis. Lexington,MA: D.C. Heath and Company.Fama, Eugene and Michael Jensen. 1983. “Separation of Ownership and Control,” Journal of Law and Economics,26: 301-326.Galbraith, Jay R. 1977. Organization Design. Reading, Massachusetts: Addison-Wesley.Galbraith, Jay R. 1995. Designing Organizations: An Executive Briefing on Strategy, Structure, and Process.San Francisco: Jossey-Bass.Garicano, Luis. 2000. “Hierarchies and the Organization of Knowledge in Production.” Journal of PoliticalEconomy 108(5): 874-904.23Geankoplis, John and Paul Milgrom. 1991. “A Theory of Hierarchies Based on Limited Management Attention.”Journal of the Japanese and International Economies 5(3): 205-225.Guthrie, James P. forthcoming. “High Involvement Work Practices, Turnover and Productivity: Evidence fromNew Zealand.” Academy of Management Journal.Hackman, J. Richard and Gene R. Oldham. 1980. Work Redesign. Reading, MA: Addison-Wesley.Hackman, J. Richard and Edward E. Lawler, III. 1971. “Employee Reactions to Job Characteristics.” Journal ofApplied Psychology 55(3): 256-286.Hayek, Friedrick von. 1945. “The Use of Knowledge in Society,” American Economic Review 35(4): 519-530.Hom, Peter W., Fanny Caranikas-Walker, Gregory E. Prussia, and Rodger W. Giffeth. 1992. “A Meta-AnalyticalStructural Equations Analysis of a Model of Employee Turnover.” Journal of Applied Psychology 77(6): 890-909.Hosseini, Hamid. 1998. “Seeking the Roots of Adam Smith’s Division of Labor in Medieval Persia.” History ofPolitical Economy 30(4): 653-681.Huselid, Mark. 1995. “The Impact of Human Resource Management Policies on Turnover, Productivity, andCorporate Financial Performance.” Academy of Management Review 38(3): 635-672.Ichniowski, Casey and Kathryn Shaw. 1995. “Old Dogs and New Tricks: Determinants of the Adoption of Productivity-Enhancing Work Practices.” Brookings Papers: Microeconomics 1-65.Ichniowski, Casey and Kathryn Shaw and Giovanna Prennushi. 1997. “The Effects of Human Resource ManagementPractices on Productivity: A Study of Steel Finishing Lines.” American Economic Review 87(3): 291-313.

Jensen, Michael and William Meckling. 1992. “Specific and General Knowledge and Organizational Structure.”In Contract Economics, ed. Lars Werin and Hans Wijkander. Oxford: Blackwell.Jensen, Michael and Karen Wruck. 1994. “Science, Specific Knowledge, and Total Quality Management.” Journalof Accounting and Economics 18(3): 247-287.Kreps, David. 1997. “Intrinsic Motivation and Extrinsic Incentives.” American Economic Review Papers andProceedings 87(2): 359-364.Lawler, Edward E. 1986. High-Involvement Management. San Francisco: Jossey-Bass.Lawler, Edward E. 2000. Rewarding Excellence: Pay Strategies for the New Economy. San Francisco: Jossey-Bass.Lawler, Edward E., Susan Albers Mohrman, and Gerald E. Ledford, Jr. 1998. Strategies for High PerformanceOrganizations–The CEO Report. San Francisco: Jossey-Bass.Lawrence, Paul and Jay Lorsch. 1967. Organization and Environment: Managing Differentiation and Integration.Boston: Harvard Business School Press.Lazear, Edward. 1992. “The Job as a Concept.” In Performance Evaluation and Incentives, ed. William Bruns,pp. 183-215. Boston: Harvard Business School Press.Lazear, Edward. 2000. “Performance Pay and Productivity.” American Economic Review 90(5): 1346-1361.Levine, David I. 1995. Reinventing the Workplace: How Business and Employees Can Both Win. Washington,D.C.: Brookings Institution.Lindbeck, Assar and Dennis Snower. 2000. “Multi-task Learning and the Reorganization of Work: From Tayloristicto Holistic Organizations. Journal of Labor Economics 18(3): 353-376.MacDuffie, John Paul. 1995. “Human Resource Bundles and Manufacturing Performance: Organizational Logicand Flexible Production Systems in the World Auto Industry.” Industrial and Labor Relations Review 48: 197-221.24Medoff, James and Katherine Abraham. 1980. “Experience, Performance and Earnings.” Quarterly Journal ofEconomics. XCV:702-736.Milgrom, Paul and John Roberts. 1990. “The Economics of Modern Manufacturing: Technology, Strategy, andOrganization.” American Economic Review. 80(3): 511-528.Milgrom, Paul and John Roberts. 1995. “Complementarities and Fit: Strategy, Structure, and OrganizationalChange in Manufacturing,” Journal of Accounting and Economics 19(2-3): 179-208.Miller, Eric J. and A.K. Rice. 1967. Systems of Organization: The Control of Task and Sentient Boundaries.London: Tavistock Publications.Mobley, William H. 1982. Employee Turnover: Causes, Consequences, and Control. Reading, Massachusetts:Addison-Wesley.Mohrman, Susan A., Susan G. Cohen, and Allan M. Mohrman, Jr. 1995. Designing Team-Based Organizations:New Forms for Knowledge Work. San Francisco: Jossey-Bass.Moorman, Jere. 1983. The Humorous Dictionary of Economics. San Diego: Crane Publications.

Murdock, Kevin. 1998. “Intrinsic Motivation and Optimal Incentive Contracts.” Working paper, Stanford BusinessSchool.Prahalad, C.K. and Gary Hamel. 1990. “The Core Competence of the Corporation.” Harvard Business Review68(3): 779-92.Rosen, Sherwin. 1982. “Authority, Control, and the Distribution of Earnings.” Bell Journal of Economics 13(2):311-323.Rosen, Sherwin. 1983. “Specialization and Human Capital.” Journal of Labor Economics 1: 43-49.Rousseau, Denise M. 1995. Psychological Contracts in Organizations: Understanding Written and UnwrittenAgreements. Thousand Oaks, CA: Sage Publications.Ryan, Richard M. and Edward L. Deci. 2000. “Self-Determination Theory and the Facilitation of Intrinsic Motivation,Social Development, and Well-Being.” American Psychologist 55(1): 68-78.Sah, Raaj and Joseph E. Stiglitz. 1986. “The Architecture of Economic Systems: Hierarchies and Polyarchies.”American Economic Review 76(4): 716-727.Senge, Peter M. 1990. The Fifth Discipline: The Art and Practice of the Learning Organization. New York: CurrencyDoubleday.Smith, Adam. 1776. The Wealth of Nations. Reprinted by Modern Library.Steers, Richard M. and Lyman W. Porter (eds.). 1991. Motivation and Work Behavior, Fifth Edition. New York:McGraw-Hill.Stevens, Margaret. 1996. “Transferable Training and Poaching Externalities.” In Acquiring Skills: Market Failures,Their Symptoms and Policy Responses, eds. Alison L. Booth and Dennis J. Snower. Cambridge, England:University Press.Taylor, Frederick. 1923. The Principles of Scientific Management. New York: Harper and Row.Thompson, James D. 1967. Organizations in Action: Social Science Bases of Administrative Theory. New York:McGraw-Hill Book Company.Trist, Eric L. and Ken W. Bamforth. 1951. “Some Social and Psychological Consequences of the LongwallMethod of Coal-Getting.” Human Relations 4: 3-38.Whyte, William Foote, Melville Dalton, Donald Roy, Leonard Sayles, Orvis Collins, Frank Miller, GeorgeStrauss, Friedrich Fuerstenberg, and Alex Bavelas. 1955. Money and Motivation: An Analysis of Incentives inIndustry. Westport, CT: Greenwood Press Publishers.Williamson, Oliver. 1975. Markets and Hierarchies. New York: Free Press.APPENDIX

Surveys and Recent Textbooks on Personnel and Organizational EconomicsBaron, James and Michael T. Hannan. 1994. “The Impact of Economics on Contemporary Sociology.” Journalof Economic Literature 32(3): 1111-1146.Baron, James and David Kreps. 1999. Strategic Human Resources: Frameworks for General Managers. NewYork: Wiley.Brickley, James, Clifford Smith and Jerold Zimmerman. 2001. Organizational Architecture: A Managerial EconomicsApproach. New York: McGraw Hill-Irwin, second edition.

Gibbons, Robert. 1999. “Taking Coase Seriously,” Administrative Science Quarterly 44: 145-157.Gibbons, Robert. forthcoming. “Firms (and Other Relationships).” In The Twenty-First Century Firm: ChangingEconomic Organization in International Perspective, ed. Paul DiMaggio. Princeton: Princeton University Press.Gibbons, Robert. 1998. “Game Theory and Garbage Cans: An Introduction to the Economics of Internal Organization.”In Debating Rationality: Nonrational Aspects of Organizational Decision Making, eds. Jennifer Halpernand Robert Stern. Ithaca, NY: ILR Press.Gibbons, Robert and Michael Waldman. 1999. “Careers in Organizations: Theory and Evidence.” In Handbookof Labor Economics v.3B, eds. Orley Ashenfelter and David Card. Amsterdam: North Holland.Hirsch, Paul, Stuart Michaels and Ray Friedman. 1987. “’Dirty Hands’ Versus ‘Clean Models’: Is Sociology inDanger of Being Seduced by Economics?” Theory and Society 16: 317-336.Jensen, Michael. 1998. Foundations of Organizational Strategy. Boston: Harvard University Press.Kaufman, Bruce E. 1999. “Expending the Behavioral Foundations of Labor Economics,” Industrial RelationsReview 52(3): 361-392.Kreps, David. 1990. “Corporate Culture and Economic Theory.” In Perspectives on Positive Political Economy,eds. James Alt and Kenneth Shepsle. Cambridge: Cambridge University Press.Lazear, Edward. 1998. Personnel Economics for Managers. New York: Wiley.Lazear, Edward. 1991. “Labor Economics and the Psychology of Organizations,” Journal of Economic Perspectives5(2): 89-110.Lazear, Edward. 1999. “Personnel Economics: Past Lessons and Future Directions,” Presidential Address to theSociety of Labor Economics, San Francisco, May 1, 1998. Journal of Labor Economics 17(2): 199-236.Merchant, Kenneth, Wim Van der Stede, and Liu Zheng. 2000. “Disciplinary Constraints on the Advancement ofKnowledge: The Case of Organizational Incentive Systems.” Working paper, Marshall School of Business, Universityof Southern California.Milgrom, Paul and John Roberts. 1992. Economics, Organization and Management. New York: Prentice Hall.Prendergast, Canice. 1998. “What Happens Within Firms?” In Labor Statistics Measurement Issues, eds. JohnHaltiwanger, Marilyn Manser, and Robert Topel. Chicago: University of Chicago Press for NBER.Prendergast, Canice. 1999. “The Provision of Incentives in Firms,” Journal of Economic Literature 37(1): 7-63.Simon, Herbert. 1991. “Organizations and Markets,” Journal of Economic Perspectives 5(2): 25-44.Winship, Christopher and Sherwin Rosen. 1988. “Sociological and Economic Approaches to the Analysis ofSocial Structure,” American Journal of Sociology 94: S1-S16.OrganizationalDesignJob Design &Decision Making- task bundling- hierarchy & decisionmaking- organizational units

Internal Labor Markets- recruitment- training- promotion- retention/ turnoverPerformance Evaluation& Incentives- performance measurement- subjective evaluation- pay for performanceImplicit Contracting- participation- communication systems- unionsBusinessOutcomesFinancial- profit- cost- stock price- market share- growthNon-financial- quality- service- variety- timelinessFuture- innovation- adaptabilityStrategyProduct Design- cost- quality- service- variety- timelinessProduction- complexity- technologyNature of Market- competition- cost pressures- suppliers- variance & growthof demandIntermediateOutcomesInvestment- skills of recruits- firm-specific & generalhuman capital- culture- employee expectationsMatching- skills tasks- tasks information- information decisionrights

Motivation- involvement- incentives- coordinationEnvironmentLabor Market- demographicsLegal- employment laws- regulationsSocial & PoliticalFigure 1. Systems View of Organizational Design (based on Beer, et al)27

Core Job characteristics Critical Psychological States OutcomesSkill VarietyTask Identity Experienced Meaningfulness of the WorkTask SignificanceHigh Internal Work MotivationAutonomy Experienced Responsibility for Outcomes of the WorkFeedback Knowledge of the Actual Results of the Work ActivitiesGrowth Need StrengthModerators: Knowledge & SkillContext SatisfactionFigure 2. Hackman-Oldham Model of Intrinsic Motivation Through Job Design1. Technological Requirements Technical Interdependence: Low HighLow Traditional job design Traditional group designTask Uncertainty:High Job enrichment Self-regulating teams2. Social / Psychological Social Needs:Requirements Low HighGrowth Need Low Traditional job design Traditional group designStrength:High Job enrichment Self-regulating teamsTable 1. Determinants of Emphasis on Job Enrichment or Team Production (Lawler)