pembangunan kapasitas dan kelembagaan sektor publik

Measuring Government Management Capacity:A Comparative Analysis of

City Human Resources Management Systems

Amy Kneedler Donahue, Sally Coleman Selden,and Patricia W. Ingraham

Syracuse University

Originally prepared for presentation atthe fifth national Public ManagementResearch Conference at College Station,Texas, December 3-4, 1999.

J-PART 10(2000):2:381-411

ABSTRACT

This article fills a gap in both the public management andhuman resources literatures by applying a conceptual modelsupported by a criteria-based evaluative framework to assessand compare the nature and capacity of city government human

. resources management systems. Various management reformshave swept through many American governments recently, butpractitioners and researchers have not reflected carefully on howthese reforms contribute to management effectiveness. Onemanagement system that has received relatively little systematicattention is human resources management. The easting researchabout assessing human resources is sparse, focuses on the privatesector, and fails to converge upon a set of criteria for evaluatinghuman resources management systems comprehensively. In ear-lier work, we proposed a theory that dissects the black box ofgovernment management to identify key management systems anddefine their contribution to management capacity and to overallgovernment performance. In this article, we refine this model bydeveloping a set of criteria that serve as indicators of the effec-tiveness of human resources management systems. We apply ourframework and criteria to a sample of cities in an empiricalanalysis that measures human resources management capacityand controls for two key environmental contingencies: unioniza-tion and government structure. We find that higher capacitygovernments are able to achieve better human resources out-comes, and that more unionized governments and those thata senior professional administrative officer generally have lowerhuman resources management capacity.

How governments translate resources into services is aquestion of long-standing significance to public administration,

381 /Journal of Public Administration Research and Theory


political science, and public policy analysis. Analysis of this issuemost often has focused on contingent influences on service deliv-ery: level of economic development, partisan influence, and polit-ical/administrative structural arrangements. This article is anexploratory examination of another set of influences that are cen-tral to the question of performance: management systems andmanagement capacity. Government management systems andcapacity are key components of the blade box of public adminis-tration; they comprise a set of intervening variables in the equa-tion that relates public production inputs and policy and programperformance.

In this article, we apply a model of government managementdeveloped in our earlier theoretical work to examine empiricallyvariation in management systems and capacity across govern-ments. Specifically, we evaluate human resources managementsystems in twenty-nine of the largest cities by revenue in theUnited States,1 and we relate the capacity of these systems tospecific human resources management outcomes. We alsoaccount for the extent to which two important environmentalcontingenciesdifferences in forms of urban government andlevel of unionizationinfluence human resources management.

We have chosen to pursue analysis of one managementsystemhuman resources managementfor two reasons. First,this area of government management has been a target of manysignificant reform efforts, beginning with the 1883 Pendleton Actthat first established the modern civil service and continuingthrough the contemporary downsizing and streamlining efforts ofthe National Performance Review and its local-level reinventioncounterparts (Ingraham 1995; Kettl et al. 1996). Insights into thenature of and impacts on human resources management thus cap-ture the interest of researchers, politicians, and public managersalike. Second, while human resources management functionsreceive considerable attention in the literature, as we will illus-trate below, careful explication of how to measure and evaluatepublic human resources management systems is lacking and pres-ents a compelling research opportunity. On this basis, we beginto address five primary research questions in this analysis:

How do major city governments in the United States vary interms of their human resources management systems andcapacity?

Is variation in human resource capacity associated with'Based on U.S. Census figures for variation in critical environmental contingencies facing cityFY 1995. governments?

3&2/J-PART, April 2000


Is variation in human resource capacity associated withvariation in the outcomes of human resources managementsystems in the cities?

Is variation in critical environmental contingencies asso-ciated with variation in the outcomes of human resourcesmanagement systems in the cities?

What do these findings suggest about the linkage betweenmanagement capacity and government performancegenerally?

The empirical work presented here is exploratory. It is afirst attempt to survey the characteristics of human resourcesmanagement systems comparatively, to quantify the level ofmanagement capacity that governments derive from these sys-tems, to control for the political context, and to measuremanagement outcomes. We are able to test hypotheses about theinfluences of capacity, government structure, and labor-management relations on some human resources management out-comes in a rough way and to draw some preliminary conclusions.In short, our findings demonstrate that management capacity canbe quantified and that it does have an independent impact onmanagement effectiveness. This provides some insight into twomodern reform trends: the movement from patronage to modemmanagement systems and the transformation of rigid union con-tracts into more flexible partnerships. It also permits us todevelop a future research agenda that homes in on the relation-ship between government management systems, the politicalenvironment, the quality of government management, andultimatelypolicy performance, more directly and substantially.

The structure of the article is as follows: We first brieflypresent our earlier theoretical work and describe the largerresearch project that serves as the empirical context for thisstudy. We next review the literature in which this work isgrounded. Then we describe the framework for this analysis,state our hypotheses, and explain our research methods. Finallywe present and discuss our findings, draw some preliminaryinferences, and conclude with an outline of future research direc-tions.

THEORETICAL AND EMPIRICAL SETTING

Comparative assessment of governments is founded in thepolicy analysis literature. Early efforts focused at length on theextent to which socioeconomic conditions, the development ofpolitical institutions, and the party competition and control

3S3/J-PART, April 2000


*We first presented our theoreticalframework for examining governmentmanagement at the fourth national PublicManagement Research Conference inNovember 1997. This early model isforthcoming in Jeffery L. Brudney,Laurence J. O'Toole, and Hal Rainey,eds. Advancing Public Management: Newdevelopments in Theory, Methods, andPractice. Washington, D.C.: GeorgetownUniversity Press. We have since refinedour conceptualization of governmentmanagement, as presented in a chapterforthcoming in Laurence E. Lynn Jr., ed.Models and Methods far the EmpiricalStudy of Governance. Washington, D.C.:Georgetown University Press.

contributed to different policy characteristics and general level ofprogram delivery (Gray 1973; Hofferbert 1966; lineberry andFowler 1967; Walker 1969). Intergovernmental influences onresources, the scope of administrative discretion and influence,and administrative reform also have received significant attention(see, for example, Bowling and Wright 1998; Schneider 1988;Carnevale 1995). Finally, more recent analysis addresses howadministrative capacities and structures relate to measurabledifferences in public programs and policies (Lynn, Heinrich, andHill 1999; Milward and Provan 1998; Sandfort 1999).

Although the specific characteristics or qualities of admin-istration that are identified in these later analyses of admin-istrative impact have varied, there is growing agreement thatinfluences associated with administrative arrangements do matterto the efficacy of the policy and program delivery system. Inmany respects this new consensus has common sense appeal, butits late emergence attests to the daunting methodological chal-lenges of specifying and measuring the complex dimensionality ofgovernment administration and the multiple and interrelatedinfluences it is likely to have. Further, assessing the potentialimpact of the administrative black box on government perform-ance requires that we formally acknowledge the profound influ-ence administrative activities have in shaping policy and programobjectives and outcomes.

As we have explained in earlier work, we propose that oneapproach to dissecting the government black box and understand-ing its operation is through the analysis of the infrastructure ofpublic administrationwhich we call management systems. Thegoal of our broad research agenda has been to develop a compre-hensive and valid evaluation of government management that atonce supports fruitful academic study, effectively communicatesthe nature and results of government management systems to citi-zens, and assists public managers to understand and learn aboutsuccessful management practices. In 1997, we initiated this pro-cess by proposing a preliminary model of government manage-ment performance. Later, we further developed this early modelinto a conceptual framework that can support comparative empir-ical analysis.2

We have argued elsewhere that government's administrativesystems can contribute to an overall dimension of managementcapacity. We have defined this capacity as government's abilityto develop, direct, and control its resources to support the dis-charge of its policy and program responsibilities (Ingraham andKneedler 2000b). A government's ability to marshal its resourcesis housed within its core administrative functions; it involves

384/J-PARr, April 2000


generic staff activities such as financial management, humanresources management, capital management, and information tech-nology management. These activities, immersed in a context richin political exigencies, interact in highly complex ways to influ-ence a government's performance.

Management capacity, in turn, is a necessary antecedent toeffectiveness in government organizations because it shapes andsupports longer-term performance capabilities (Ingraham andKneedler 2000b). We have argued that management capacity isitself forged not only by the qualities of the various managementsystems it encompasses, but by the absence or presence of inte-gration across the systems and by the absence or presence of asystem of managing for results. The presence of both results-oriented management and integration will, we argue, optimallyorient government capacity toward the pursuit of specific goalsand objectives. Exhibit 1 shows our model of governmentmanagement capacity.

This model undergirds a major research initiative of theAlan K. Campbell Public Affairs Institute at the Maxwell Schoolof Citizenship and Public Affairs at Syracuse University, knownas the Government Performance Project (GPP). The GPP is afive-year effort, funded by the Pew Charitable Trusts, to ratethe quality of management in state and local governments andselected federal agencies in the four areas depicted in exhibit 1:financial management, human resources management, capitalmanagement, and information technology management. Toaccomplish as comprehensive an evaluation as possible, the GPPrelies on a multimethod data gathering effort that includes asubstantial written survey, collection of archival documentation,and extensive follow-up interviews with government actors andexternal stakeholders.

The GPP analysis depends fundamentally on criteria-basedassessment. The use of this type of assessment approach is welldocumented and supported in the evaluation literature (Patton andSawicki 1986; Easton 1973; Rossi and Freeman 1989; Weimerand Vining 1992; Kaplan and Norton 1996), and it also forms thebasis of several well-known and accepted government assessmentschemes (see, in particular, the Baldrige National QualityProgram and the President's Quality Award Program). Like otherreport-card driven schemes, our use of criteria explicitly focusesdata collection and analysis efforts around stated normativeassertions about the characteristics of good management. Bydeveloping and applying a scheme of criteria that represents thedesired characteristics of government management systems andthe various functions and activities they comprise, we identify the



Exhibit 1The Overall Government Management System

PolicyOutcomes

EnvironmentalContingencies

GovernmentPerformance

ManagementCapacity

Managingfor

Results rVFinancial

ManagementHuman

ResourcesManagement

Integrating Forces

Management

CapitalManagement

InformationTechnology

Management

1Subsystems

particular strengths and weaknesses of each government'smanagement systems and the degree to which they affect thegovernment's overall ability to manage. In short, we have chosencriteria that characterize a state of high management capacity; wealso recognize that any of a wide array of managerial tactics canbe successfully applied to achieve high capacity levels.

THE STATE OF PUBLIC HUMAN RESOURCESMANAGEMENT LITERATURE

Given this theoretical backdrop, we now set the stage foranalysis of our research questions by briefly describing the state

386/J-PART, April 2000


of knowledge about human resources management systems. Wewill discuss the literature about personnel functions first, andthen we will characterize key aspects of the public personnelenvironment that have important impacts on these functions. Weposit that two contingent influences are critical: the politicalstructure of urban governments and the extent to which localpublic workforces are unionized.

Organizations boil down to people; thus the ability to obtainand retain critical human skills and talentthe essence of humanresources managementis fundamental to effective managementand to organizational performance more broadly defined.Research that focuses on the linkage between public humanresources management systems and government performance issporadic and limited, however. One explanation for the paucityof work in this area is that public human resources systems havetypically been equated with civil service systems as separate fromexecutive managers or leaders. Moreover, the uniform, bureau-cratic organizational structure of civil service often is viewed ascounter to the prime concerns of senior managers and leaders:strategic management and long-term performance. As Perry andMesch (1997, 24) point out, "The managers perceive personnelspecialists as obstructionists, and personnelists perceive linemanagers as uncooperative and uninterested." Thus public per-sonnel systems are traditionally viewed as obstacles to, ratherthan enablers of, government performance.

Furthermore, perspectives on the role of human resourcesmanagement systems in the context of city government varysharply. Saltzstein (1995, 50), for example, argues that "[a]creative role for the personnel administrator is one that infusesthe organization with the values of the personnel professions. Inlocal government, personnel values are not by nature importantones . . . their natural constituency may well be less powerfulthan others. . . ."By contrast, a major city responding tothe 1999 Government Performance Project survey observed thatu[t]he Human Resources Department is shifting to a leadershipand consultative role . . . and (strategically) partnering with citymanagement." This is congruent with the arguments of Perry andMesch (1997, 25), who also characterize emergent human resour-ces management practices and systems as "strategic" and notethat a strategic human resource model places emphasis not onpersonnel values, but on ". . . tracking performance againstmission related goals in areas such as product quality, servicedelivery, and customer satisfaction."

In the face of this array of views, empirical research thatevaluates specific human resources management functions has



received considerable attention in recent years (Fitz-enz 1994;Collins 1997; Martinez 1996; Delaney and Huselid 19%;Caudron 1999; Gooden 1998; Laabs 1996; Roberts 1996), andmany studies have focused on the organizational-level impact ofhuman resources management practices (Martinez 1996; Marko-wich 1995; Davidson 1998; Ulrich 1997; Fitz-enz 1994). Theliterature also includes studies that focus on the performanceeffects of specific human resources management functions, suchas training (Knoke and Kalleberg 1994) and selection (Huselid1995). Others have examined the influence of a collection ofhuman resources management practices on organizational out-comes (Delaney and Huselid 1996; Huselid 1995; Huselid andBecker 1996). Notably, however, much of this work has been setin the private sector.

Oddly enough, human resources management has rarelybeen examined as a complete system in the context of its links toan organization's overall potential to perform, though somescholars have begun to move in this direction (Tsui and Gomez-Mejia 1988; Huselid 1993; Ulrich 1997). Huselid, for example,attempts to characterize human resources management systemsmore comprehensively, rather than relying on a functional anal-ysis. Straus (1993) introduced a framework for evaluating theeffectiveness of public personnel departments which incorporatesthe perspectives of multiple constituent groups. Straus's approachallows personnel departments to assess how effectively they areserving each constituent group in order to compare their effec-tiveness to other personnel departments and to estimate theireffectiveness over time. Finally, a recent General AccountingOffice report (1999) presents a human capital self-assessmentchecklist for federal agencies.

Despite these important steps, however, the literature lackscareful explication of how to measure and evaluate public humanresources management as an assemblage of key functions thatcontributes to an agency's management effectiveness and has animpact on government performance. This article begins toaddress these shortcomings by explicitly viewing human resour-ces management as a component system of a government's largeradministrative superstructure, by developing a comprehensiveindex of a government's human resources management capacity,and by measuring specific outcomes of the human resourcesmanagement system. Once forged, these links will form the chainthat connects people with performance; thus they will advanceour understanding of the black box of government administration.

388//-PARr, April 2000


'Other scholars pursue similar reasoning.In particular, our thanks to Kenneth J.Meier, George Frederickson, Gary John-son, and ICMA's Michele Frisby for theirinsights about classifying governments.Frisby noted that practical conventiondoes distinguish city governments but isbased primarily on anecdotes and casestudies and not on clear criteria orsystematic data. Frederickson and John-son (1999) do establish a conceptual con-tinuum ranging from political to admin-istrative governments, and they identifyan emerging form, which they termadapted, characterized by a symbioticcoexistence of political and administrativecomponents. In addition, Hansell (1999)defines four variations on the mayor-council-manager form of government.Finally, for a detailed discussion of thehistory of machine politics and reform,see especially Ross and Levine 1996.

Environmental Context of Human Resources Management:Form of Government

We recognize that public human resources management doesnot operate in a vacuum, and we cannot fully comprehend thelinkages between human capital, public management, and govern-ment effectiveness without considering the complex environmentwithin which the human resources management system operates.We submit that, particularly at the local level, one especiallyinfluential contextual factor is the structure of a government'sleadership. Research about the influence of leadership on humanresources management systems is sparse, however, and preciselyhow different structures may relate to different managementsystems and capacity is open to considerable debate.

Scholars and practitioners have yet to categorize clearlylocal government structures with precise definitions, or even witha common framework. Much of what is perceived about citygovernments is anecdotal, inconsistent, and not founded in syste-matic empirical work. In general, though, we view local govern-ments as arrayed along a continuum according to the extent towhich they have institutionalized professional management.3Anchoring one end of the spectrum are traditional, patronage-driven governments, typically associated with pre-Progressive eramachine politics. Here may be found many of today's strong-mayor forms of government. At the other extreme are govern-ments administered by professional city managers, often referredto as reformed governments. This government form was createdto move away from the machine politics model, and movementfrom partisan hiring to neutral merit systems was one part of thereform agenda. Between these opposing poles range governmentsthat have adopted professional management to varying degrees.Generally, these governments are led by a mayor with the assist-ance of a chief appointed official (CAO) who serves as a senioradministrator and whose role is to enhance the flexibility of thegovernment's administrative frameworkmost particularly itspersonnel structureto enable the government to meet its stra-tegic objectives more efficiently and effectively.

In reality, this conceptual continuum masks much of thecomplexity of the political environment. Machine politics hadeffectively defined who was hired, who was promoted, and whowas not eligible for public employment. While reformed govern-ments were intended to replace these systems, history shows thatnew merit systems often had only a modest impact on hiring,firing, and promotion. Merit systems were not well funded orstaffed, and although the total number of governments with meritsystems continued to expand until the 1930s, the actual scope of



the systems was limited. As Dresang (1999, 28) notes, formaladoption of a merit system was sometimes " . . . little more thana symbolic gesture, and old hiring and promotion systems quicklyfound ways around the new."

Further, the reformed governments' emphasis on profes-sional management suggested better support of human resourceneeds than civil service systems actually provided, and the goalsof merit were often ignored. As Nigro and Nigro (1994, 23)point out, "In many localities and states, the spoils systems werestill deeply entrenched . . . little attention was paid to potentialconnections between personnel practices and organizational pro-ductivity and effectiveness." Thus the very civil service struc-tures created to sustain merit were frequently as unwieldy as thepatronage systems they supplemented or supplanted. In fact,some more traditional strong-mayor cities, such as Indianapolis,have successfully emphasized the connection between humancapital, streamlined management systems, and better perform-ance, suggesting that the separations and distinctions betweenpolitical environment and human resources management systemsmay be becoming less clear.

Although the literature has yet to sort out the nature of citygovernment leadership and its relationship to human resourcesmanagement, we suggest that each point on the structure ofgovernment continuum is likely to have a different relationship tothe acquisition and retention of human resources. In general, wewould expect that because critical skills are not a prerequisitefor patronage appointments, more patronage-based systems wouldhave lower workforce capacity, on average, than those with civilservice systems administered by professional public managers. Todisentangle and quantify the array of possible government formsis well beyond the scope of this research effort. Nonetheless,we can make some rudimentary distinctions between governmentsthat appear to have institutionalized professional management andthose that do not, to help control for the nature of politicalenvironment in our analysis.

Environmental Context of Human Resources Management:Unionization

The nature and level of unionization and the extent ofcollective bargaining are also likely to have an impact on humanresources management systems and capacity. Comparative analy-sis of the relationships between levels of unionization andgovernment performance and productivity are limited. We doknow, however, that collective bargaining has become well estab-lished as a component of all levels of government (Goldfield



1989; Reeves 1997). While estimates of membership vary, theWinter Commission reported that about 38 percent of all publicemployees and approximately 53 percent of municipal employeeswere union members at the end of the 1980s (Ban and Riccucci1993, 79-82). Moreover, empirical evidence from local schooldistricts and fire services suggests that high levels of unionizationconstrain both flexibility and productivity (Babcock and Engberg1997; Smith and Lyons 1980). This perception underpins manycurrent labor-management partnership reforms in reinventinggovernment and elsewhere (Douglas 1992; Kearney and Hays1994; Hays and Kearney 1995).

The rigidity and complexity produced by bureaucratic civilservice systems with well-developed union clout will be, weargue, a significant constraint on the ability of top leaders toacquire and use the flexibility necessary to link human resourcesmanagement to clear performance objectives. The highest degreeof managerial autonomy and the strongest performance orienta-tion are likely to be present, therefore, either in cities withcollaborative union relationships or in cities in a nonunionizedenvironment (Goldsmith 1998).

ANALYTICAL FRAMEWORK AND HYPOTHESES

We have argued that a government's human resources man-agement system has independent effects on its performance, andthus it should be distinguished and examined in order to developa richer understanding of a government's management effective-ness. We also have acknowledged that environmental contingen-cies can powerfully affect the ability of a government's adminis-trative functions to operate effectively and efficiently, and wehave identified and described the state of knowledge about twoimportant contingencies that are likely to affect human resourcesmanagement. We now apply the general framework described inexhibit 1 to evaluate the ability of public entities to effectivelyacquire, sustain, maintain, and deploy a workforce in circum-stances shaped by structural reforms and labor-managementrelations.

As we have said, the core of our conceptual approach ismanagement capacity, such that a government's human resourcesmanagement capacity fundamentally depends on the configura-tion, procedures, and work processes of the government's humanresources management systems. Our definition of human resour-ces management capacity is also intertemporalthat is, it dependson the extent to which a government can maintain a reliableand appropriately configured base of human capital over time,success at which necessitates functions such as strategic

391IJ-PART, April 2000


workforce planning. These dimensions of human resourcesmanagement capacity, and the relationships between them, areshown in exhibit 2, which serves as the basis for our empiricalanalysis.

Hypotheses

Given this theoretical setting and our research questions, weposit five hypotheses about human resources management in citygovernments:

Hypothesis 1: Human resources management capacity is posi-tively correlated with human resources manage-ment outcomes.

Hypothesis 2: The level of unionization is negatively correlatedwith human resources management capacity.

Hypothesis 3: More traditional strong mayor governments willhave less human resources management capacity.

Hypothesis 4: The level of unionization is negatively correlatedwith better human resources management out-comes.

Hypothesis 5: More traditional strong mayor governments willhave poorer human resources management out-comes.

The exploratory analysis in this article begins the complex andchallenging task of examining these hypotheses by evaluatingwhether a government is able to locate the right people in theright places at the right time to meet its administrative needs.Our assessment is based on a set of criteria that define thecharacter of a government's human resources management systemand indicate the extent to which it is effective. We will nowpresent the development and substance of these criteria.

Human Resources Management Assessment Criteria

An important component of our approach throughout theGovernment Performance Project has been to define the criteriaby which we assess governments. To accomplish this, we con-vened a broad group of practitioners and scholars who are con-sidered to be experts in each area of government management ofconcern to the GPP, and we asked them what they consider to bethe most significant elements of these management systems,based on the literature of the field, their experience, and their



Exhibit 2The Human Resources Management System

EnvironmentalContingencies

HumanResources

ManagementOutcomes

HumanResources

ManagementCapacity

Components of Human ResourcesManagement Capacity

own research. Through an iterative series of meetings and writtenfeedback, we discovered a high degree of consensus about whatmattered most to successful government management, and weconverted these findings into evaluation criteria that met theapproval of the entire group.

This study uses five criteria to characterize sound humanresources management in city governments. The criteria describe



what are generally held to be the vital components of successfulsystems, generally capturing recruiting, hiring, retaining, promot-ing, appraising, compensating, motivating, training, and terminat-ing public employees. They also reflect an emerging, broad con-sensus in the literature and among our expert advisors about thecritical levers of effectiveness in human resources management(Perry and Mesch 1997; National Institute of Standards andTechnology 1999; OPM 1999; USGAO 1999). In particular, wefound widespread agreement that key components of good humanresources management systems include the use of coherent rulesand procedures; efforts at workforce planning; the ability tofacilitate timely hiring; sophisticated professional developmentprograms; and meaningful reward structures and disciplinary pro-cedures. In addition, because increased flexibility, discretion, anddelegation of authority in the human resources management pro-cess have been a consistent focus of administrative reform, manyagreed that it is also important to consider where in the systemand for whom such flexibility exists. Finally, as we have alreadynoted, contextual factorssuch as the relationship between polit-ical appointees and career civil servants within governments andthe disposition of labor-management relationswere seen to havesignificant impacts on the ability of governments to marshal theirhuman capital effectively. Our criteria ultimately were presentedto the National Association of State Personnel Executives, whichconcurred with their definition.

To facilitate data gathering and analysis, we consolidated theviewpoints of our advisors and reviewers into the following fivecriteria:

Criterion 1: Planning for the Workforce. This criterionhinges on the extent to which a government is aware of andaddresses its personnel capacity over time, particularly thesophistication with which the government conducts strategicanalysis of present and future human resource needs and avail-ability. Such foresight depends on three key activities. First, thegovernment must collect sufficient data about its workforce tosupport evaluation of its current status and projections of futurerequirements. Second, the government must actively engage incomprehensive needs analysis. Third, the government mustdevelop detailed action and contingency plans to meet the futureworkforce needs it identifies. An outcome associated directlywith this criterion is whether the government has adequateinformation about its workforce to plan effectively.

Criterion 2: Hiring the Workforce. This criterion addressesthe extent to which the government is able to obtain theemployees it needs. To accomplish this, a government must be



able to conduct effective recruiting efforts and to hire appro-priately skilled and qualified employees. An outcome associateddirectly with this criterion is the ability hire employees in atimely manner. Substantial evidence suggests that obstacles totimely hiring of qualified public employees, such as the inabilityto compete with private-sector salaries and benefits and complexprocedural requirements, drastically inhibit the efficiency ofpublic agencies. There is also wide agreement that the managerswithin the government's subunits are likely to know the mostabout what kind of staff that unit needs, and thus these managersmust have discretion in the hiring process.

Criterion 3: Sustaining the Workforce. This criterion con-cerns the government's ability to maintain an appropriatelyskilled workforce by conducting or providing training to developand maintain employee skills, by retaining skilled and experi-enced employees, by disciplining poor performers, and by termi-nating employees who cannot or will not meet performance stan-dards. In this case, governments must invest in programs thatdirectly address employee performance, and focus on improvingit or bringing it in line with organizational objectives. In addi-tion, this criterion suggests that a government should separate anemployee as soon as it is evident that he or she is not capable ofcontributing to overall performance of the government. An out-come based on this criterion is the ability of a government toterminate an employee in a timely manner.

Criterion 4: Motivating the Workforce. This criterionfocuses on whether a government is able to encourage employeesto perform effectively in support of the government's goals.Effective motivation typically rests on the use of appropriatemonetary and nonmonetary rewards and incentives, an effectiveperformance appraisal system, and sound mechanisms that facili-tate employee feedback. Governments are oft cited as laggardswith respect to the adoption of effective motivational devices,particularly with respect to pay and benefits, leaving themencumbered by problems of complacency and poor morale.Those that can overcome these challenges can be expected toachieve their performance goals more readily.

Criterion 5: Structuring the Workforce. This criterion cap-tures the degree to which the government's human resourcesstructure supports its ability to achieve its workforce goals. Thisincludes having a coherent and appropriately sized classificationssystem reinforced by personnel policies that are flexible in termsof promotion and compensation. While there is some debate inthe public human resources field about what structural forms arebest, recent reform trends emphasize flexibility and performance.


"We would like to thank Yilin Hou,Willow Jacobsen, EUen Rubin, andJessica Sowa for their research assistance.

5As suggested by Larsson (1993, 1532),the designer of the coding scheme wasnot used to code the responses. FiveSyracuse University graduate studentswere hired to code the original surveyresponses. The designer of the codingscheme briefed each rater and providedgrading instructions.


In particular, there is clear movement, both in the United Statesand internationally, toward simplification of personnel rules andprocedures and increased use of provisional workers.

DATA, MEASURES, AND METHODS4

The hypotheses are examined using data from the Govern-ment Performance Project. In March 1999, the GPP administeredto die thirty-five largest cities (by FY 1995 revenue) in theUnited States a survey that included a section about humanresources management practices. At the time of our analysis,twenty-nine cities had completed and returned surveys for aresponse rate of approximately 82.9 percent (a list of respondentcities is presented in exhibit 3). The human resources manage-ment section of the survey instrument contained twenty-fourmultipart open-ended questions designed to yield informationabout a given city's status with respect to each of the criteriadescribed above. The survey instrument was pretested in fourstates, four local governments, and four federal agencies in 1997.Based on this pilot study, the instrument was revised, streamlinedto focus as directly as possible on the evaluation criteria, andcustomized to each level of government. After completing asurvey of fifty states in 1998, the survey was again revised.

Since the survey instrument consisted of open-ended ques-tions, the following procedure was used to code the data (Larsson1993):

A coding scheme was designed for systematic conversion ofthe qualitative survey responses and supporting city docu-mentation (see appendix 1) into quantified variables.

The coding scheme was pretested, using the surveyresponses and supporting documentation of five cities.

The coding scheme was revised, reflecting feedback fromthe pretest.

The survey responses were coded, using two raters for eachcriteria.3

Inter-rater reliability was computed.

A process for resolving coding discrepancies was developedand employed.

We checked interrater reliability by using two techniques. First,we computed the pairwise percentage agreement (Larsson 1993).



Exhibit 3Human Resources Management Capacity Scores by City

City

AustinBaltimoreBostonBuffaloChicagoClevelandColumbusDallasDenverDetroitHoustonIndianapolisJacksonvilleKansas CityLos Angeles

CapacityScore

582919142222242746462245323129

Chy

MemphisMilwaukeeMinneapolisNashvilleNew OrleansNew YorkPhiladelphiaPhoenixRichmondSan AntonioSan JoseSeattleVirginia BeachWashington, D.C.Mean = 34.4; SD

CapacityScore

2439534020424865234526393235

= 12.73

'Larsson (1993) recommends a consensusapproach as opposed to an averagingapproach to resolve coding discrepancies.

7A weighting scheme is necessary toexplain why a government's humanresources capacity depends unevenly onthe characteristics of various humanresources management functionsin otherwords, assessment of a government'smanagement capacity ought to addressboth the government's managerialstrengths and weaknesses and the relativeimpact or importance of these strengthsand weaknesses for capacity. Forexample, it may be more important thatthe government plan ahead to meet itsfuture workforce needs than that thegovernment provide sufficient opportuni-ties for employee feedback, though bothfactors contribute to human resourcesmanagement capacity and ultimately to agovernment's overall management effec-tiveness.

In total, we had 6,760 observations evaluated by two raters. Theraters coded responses alike for 6,561 observations for a pairwisepercent agreement of 97.06. Second, we computed the correlationbetween the two coders' ratings. The correlation is 0.96, whichsuggests a great degree of consistency between the raters. Toresolve the 199 discrepancies that arose between raters, a Syra-cuse University graduate student and the designer of the codingscheme reexamined the observations, discussed the responses,conducted follow-up interviews with city officials where neces-sary, and reached a joint consensus about how to code them(Larsson 1993, 1532).6

The GPP survey was designed to assess five criteria. Animportant issue when criteria-based assessment is used is thechoice of a coherent structure for the system of evaluationindicators that assigns them appropriate weights.7 In this study,the human resources management variables we coded weregrouped by criteria, were explicitly assigned weights, and weresummed to create an index per criteria and an overall capacityindex (DeVellis 1991). We used the following approach toconstruct the indices:

We linked the quantified human resources variables to eachcriterion (see appendix 2).


*A detailed summary of the employedweighting schemes is available by requestfrom the authors. Carmines and Zeller(1979) indicate that it is desirable to havean alpha coefficient of .70 or more. How-ever, DeVellis (1991, 85) views alphacoefficients between .60 and .70 asacceptable, though not desirable.Although criteria 2, 4, and 5 have alphasfalling within this range, they are rela-tively stable when items are added orremoved (DeVellis 1991). This is particu-larly important in our study given therelatively small sample size (DeVellis1991,86).

*Out of 100 points, we assigned criteria1-4 a value of 22.5 and criterion 5 avalue of 10 (for a maximum value of100).


We assigned weights to each variable assigned to eachcriterion.8

We summed the weighted variables.

We standardized the scales.9

We computed a measure of human resources managementcapacity for each city by summing the five standardizedscales (overall Cronbach's alpha = .76).

We applied two other weighting schemes to assess thesensitivity of the human resources management capacitymeasure and determined it to be robust.

Appendix 1 presents the quantified variables associated with eachcriterion, the Cronbach's alpha for each criterion, and the meansand standard deviations of the index scores across cities.

We also included two critical environmental contingencies inthis study. The first was the extent of unionization in a city, asmeasured by the percentage of the city's workforce that iscovered by collective bargaining agreements. This informationwas reported by each city in its response to our written survey.The second was the form of government. Here we sought to dis-tinguish between more traditional, patronage-driven, strongmayor forms of government and those that have professionalmanagers in place, either as a mayor's chief appointed official(CAO) or as a city manager. Following Frederickson and John-son (1999), we assigned each city in our sample to one of threecategories:

Political (or traditional): Those cities in which the mayoracts as the chief executive officer with administrative author-ity over the executive function of the city under a mayor-council form of government.

Adapted: Those cities in which the responsibility for admin-istrative functions is shared between the mayor and a profes-sional manager.10

Administrative (or reformed): Those cities in which a pro-fessional administrator, appointed by the elected body, hasresponsibility for implementing the city's laws and policies.

'"For a case example of the adapted type . I - J T J , , . , , , , __of government, see Governing magazines C m e s w e r e Classified On the basis of four SOUTCeS of data. Therecent story on Oakland, California International City-County Management Association (ICMA) clas-(Gurwitt 2000). sified the cities in our study according to whether or not they

398/ J-PART, April 2000


have a CAO." We also examined city organizational charts,reviewed city web pages, and conducted follow-up interviews toreconcile discrepancies. We then created a city classificationcontinuum that assigned a value of 1 to political/traditionalgovernments, a value of 2 to adapted cities, and a value of 3 toadministrative/reformed governments.12 Appendix 3 presents themeasurement scale, mean, and standard deviation for both controlvariables.

Finally, the study examines four human resources manage-ment outcomes:

whether the city has the information it needs about its work-force to plan effectively;

the average time it takes a city to hire someone to fill aposition;

the average time it takes a city to terminate an employee;and

the percentage of employees who are terminated duringprobation.

This information was reported by each city in its response to ourwritten survey. Appendix 3 presents the measurement scale,mean, and standard deviation of each outcome variable.

In light of the exploratory nature of this analysis, weexamine the hypotheses enumerated above using bivariate andpartial correlations. In addition, we rate cities according to theirhuman resource capacity, and we provide descriptive statisticsabout the sample, high performers, and low performers. We willthen present and discuss our results.

FINDINGS AND DISCUSSION

"Our thanks to ICMA's Michele Frisbyfor providing this information.

''The classifications of the cities in oursample are available from the authorsupon request.

Overall, we find that the cities in our sample vary a greatdeal in terms of their human management systems and capacity.The index of human resources management capacity by city ispresented alphabetically in exhibit 3. Scores range from a high of65 to a low of 14, with a mean of 34.44 and a standard deviationof 12.73. Phoenix and Austin receive the highest capacity scores(65 and 58, respectively), while Buffalo and Boston register thelowest (14 and 19, respectively). Virginia Beach, Jacksonville,andsurprisingly, given its unique situation and anecdotal historyof poor management and system abusesWashington, D.C. areabout average in terms of human resources management capacity.



The highest capacity cities are interesting because these arethe cities that others are likely to copyhigh performers fre-quently serve as benchmarks. As Meier and Gill (2000, 1) pointout, "The supposition that public managers seek to identify andemulate the average performing case is simply wrong." Con-versely, the reason to look at poor performers is to gain a betterunderstanding of the constraints on performance and the barriersto good management that are faced by these cities. Exhibit 4compares the critical contingencies and organizational character-istics of the best performers and the poorest performers as wellas the sample of cities as a whole.

As exhibit 4 shows, the size of the workforce in the highestand the lowest performing cities is similar, on average. Thevariation of the number of classification titles is also small acrossthese cities. Turnover is almost 4 percent higher in the highestcapacity cities. Finally, New York City is an outlier in the data,with over 111,000 provisional and nonclassified employees, or 45percent of their workforce. If New York is omitted from thesample calculations, the mean percentage of provisional and non-classified employees across the cities in the sample falls to 18percent from the 26.8 percent reported in exhibit 4. Thus, we

Exhibit 4Critical Contingencies and Organizational Characteristicsof High and Low Performers

ContingenciesUnionizationAdministrative/reformedTraditional/political

CharacteristicsFY 95 revenue (dollars)1998 population estimateTotal employeesClassified employeesProvisional/nonclassifiedClassification titles1998 turnover

Cities withHighest Capacity

(Mean)

34.5%1.000.00

1,369,573875,24912,89510,579 (82%)2,316(18%)8519.28%

Cities withLowest Capacity

(Mean)

88%0.001.00

1,442,833428,08212,34110,828 (87.7%)1,513(12.3%)9005.50%

Entire Sampleof Cities(Mean)

60%0.280.48

3,478,310tl,110,073tt23,39417,126(73.2%)6.268 (26.8%)*9508.76%

fThis fells to 1,913,137 if New York city is omittedttThis fells to 884,713 if New York city is omittedfThis fells to 18% if New York city is omittedsee discussion



can see that substantially fewer provisional and nonclassifiedworkers are employed in the lowest performing cities.

In terms of the two critical contingencies we discuss andmeasure, exhibit 4 shows that, on average, 34.5 percent of thehighest performing cities' workforces are unionized, compared to88 percent of the lowest performers' workforces and 60 percentof the sample cities' workforces, on average. Additionally, bothof the highest performing cities have administrative/reformedgovernments, while the two lowest performing cities are tradi-tionaloperated by a mayor without a CAO. In the sample, 27.6percent of the cities have administrative/reformed governments,24.1 percent of the cities have adapted governments, and 48.3percent of the cities have political/traditional governments.

Examination of the highest and lowest human resourcesmanagement capacity cities suggests support for our second andthird hypotheses. Moreover, as exhibit 5 shows, more traditionaland political governments are moderately associated with loweroverall human resources management capacity. Unionization, onthe other hand, does not appear to be correlated with overallcapacity. A review of capacity disaggregated by criteria revealsthe nuances of these associations, however. Both workforceunionization and political/traditional government are significantlyassociated with less capacity to hire and to motivate city work-forces. Neither unionization nor city classification are signifi-cantly associated with sustaining or structuring the workforce orwith workforce planning.

These findings are consistent with our expectations; theliterature suggests that unions constrain managerial autonomy.

Exhibit 5Correlation of Human Resources Management CapacityCriteria and Critical Contingencies

Contingency

UnionizationCity classificationt

tClajsification is: 1 Significant at 0.10

!T!

Ove

rall

IC

apac

ity

-.04-.27*

Cri

teri

on(p

lannin

i

.09-.17

'traditional/political; 2**Significantat0.05

Cri

teri

on(h

iring

)

-.32**-.42**

f**> OX)

Crite

rion

(susta

inii

.18

.03

=adapted; 3=admini

Documents

pembangunan kapasitas dan kelembagaan sektor publik