文档分类：
在线文档经过高度压缩，下载原文更清晰。
淘豆网网友近日为您收集整理了关于IT governance and business value in the public anizations — The role of elected representatives in IT governance and its impact on IT value in U.S. state governments.pdf的文档，希望对您的工作和学习有所帮助。以下是文档介绍：IT governance and business value in the public anizations — The role of elected representatives in IT governance and its impact on IT value in U.S. state governments.pdf IT governance and business value in the public anizations — The role of elected representatives in ITgovernance and its impact on IT value in U.S. state governmentsMin-Seok Pang Fox School of Business, Temple University, 1801 Liacouras Walk, Philadelphia, PA 19122, USAa b s t r a c ta r t i c l e i n f oArticle history:Received (来源：淘豆网[/p-8664602.html])19 March 2013Received in revised form 26 December epted 27 December 2013Available online 4 January 2014Keywords:Business value of ITIT governanceU.S. state governmentCost efciencyTheory of political controlPrincipal–agent modelThis paper studies IT business value in the public anizations, to which the information systems (IS)literature so far has paid little attention. Specically, we investigate the moderating effect of IT governance onthe relations(来源：淘豆网[/p-8664602.html])hip between IT investments and government performance. Drawing upon the theory of political con-trol on bureaucracy from the political sciences literature, we hypothesize that the presence of legislative controlson IT management increases returns to IT spending, which are measured by cost efciency. Our empirical analysisin the context of U.S. state governments shows that formal establishment of a chief information ofcer (CIO) po-sition by legislatio(来源：淘豆网[/p-8664602.html])n is a key prerequisite to positive returns from IT expenditures in state governments. Also, theimpact of IT spending on state cost efciency increases when a state senate approves appointment of CIO nom-ination than when it does not. This study contributes to the IS literature by demonstrating the importance ofelected representatives as part of IT governance in the public anizations. 2014 Elsevier B.V. All rights reserved.1. IntroductionThe public a(来源：淘豆网[/p-8664602.html])nizations are large consumers of informationtechnologies (IT). Vivek Kundra, the former CIO of the U.S. federal gov-ernment, stated that it spends as much as $80 billion every year in IT[21, p. 56]. Governments in the U.S. at all levels – federal, state, andlocal levels – are strategically using IT for a variety of purposes rangingfrom maintaining operational infrastructures to delivering responsivepublic services and interacting with citizens [49].(来源：淘豆网[/p-8664602.html]) For instance, stategovernments extensively utilize IT in administrating Medicaid, thehealthcare
programs for low-e residents [54], which con-sume a substantial share of state expenditures [48]. Thus, informationsystems (IS) researchers as well as citizens, who pay taxes for IT spend-ing, may ask a
Does every dollar spent in IT helpgovernments fulll their objectives and create sufcient value for thepublic? To the best of our k(来源：淘豆网[/p-8664602.html])nowledge, however, IS researchers haveshed little light on the IT value in the government sector, compared toa sheer number of the studies at the for-prot business setting.IS researchers have been interested in anizational practices with IT investments [42]. In this researchstream, scholars aim to identify the specic context in which rms areable to realize greater value from their IT investments. A broad rangeof studies have identied a variety of fa(来源：淘豆网[/p-8664602.html])ctors such as improved humanresource practices [8], business process reengineering [63], enhancedmarketing or supply chain capabilities [7,74], innovative, responsive ITmanagement practices [31,64], and analytic capabilities [66]. Thesestudies demonstrate that rms that possess plementary fac-tors can better exploit IT resources to anizational perfor-mance and to achieve petitive advantages.In line of this research stream, considering IT governance p(来源：淘豆网[/p-8664602.html])le-mentary practice in the relationship between IT anizationalperformance, we examine the moderating effect of IT governance onthe relationship between IT spending and performance, measured bycost efciency, in U.S. state governments. In particular, this study focuseson the role of elected representatives in IT governance. Our reading of ITgovernance studies indicates that they have mainly focused on the agen-cy side — the role of CIOs, CEOs, or othe(来源：淘豆网[/p-8664602.html])r senior executives [3,28,61,62].To the best of our knowledge, most IT governance studies have notablymissed the other key part of IT governance, which the principal–agentmodel calls principals — board of directors in business rms or electedlawmakers in the public sector. Thus, we ask: what is the role of legisla-ture in IT governance in the public anizations?Drawing upon the theory of political control on bureaucracy fromthe political sciences lite(来源：淘豆网[/p-8664602.html])rature [5,11,38], we maintain that IT gover-nance is more effective when legislative branches are involved in con-trolling and monitoring IT management. Rooted in the principal–agentmodel [45], the political control theory explains how elected represen-tatives (i.e. principals) control unelected bureaucrats (i.e. agents). Incontemporary government administration, politicians do not have assufcient information and expertise in administration as agency ofcialsdo [6,40]. Thus, there exist information asymmetry and moral hazardsDecision Support Systems 59 (–285 Tel.: +1 215 204 7676.E-mail address: mins.pang@./$ – see front matter
2014 Elsevier B.V. All rights reserved.http://dx./10.1016/j.dss.Contents lists available at ScienceDirectDecision Support Systemsjournal homepage: ate/dssin this setting, in which the agents may pursue their self-interests whichare divergent from those of the principals and their constituents [39]. Aswe will discuss below, such information asymmetry and moral hazardsmay exist in IT management as well. We consider the following controlmechanisms that are available to lawmakers in supervising bureaucraticIT management — direct oversight on IT management and operation,establishment of a chief information ofcer (CIO) position by legislation,and approval of a CIO appointment. We hypothesize that the impact ofIT spending on state cost efciency is greater in the presence of thesecontrols on IT management.Our empirical analysis proceeds in two stages. First, we measure costefciency of U.S. state governments with stochastic cost frontier estima-tion [1,40]. In the second stage, we estimate the effect of IT spending oncost efciency and the moderating effects of legislative controls. Thistwo-stage estimation produces several interesting ndings. First, it re-veals that legislative establishment of a CIO position accentuates the im-pact of IT investments on cost efciency signicantly. Further, theapproval of a CIO appointment by a legislative body increases efcientreturns from IT spending as well. Finally, contrary to the predictionsfrom theory, we nd that the presence of an IT-mittee inthe legislature attenuates the impact of IT spending on cost efciency.This study contributes to the literature on IT business value and ITgovernance on several fronts. First, we expand the boundary of ITvalue research to the public anizations, an uncharteredterritory in the IS literature. We also nd that IT governance is a keymoderator in the performance effect of IT investments. Specically, weexamine the role of principals (i.e. legislatures) in IT pared to other IT governance studies that mainly focus on agents(i.e. chief executive, CIO, or other business executives) in IT manage-ment. We offer a new nding to the IS literature that the involvementof principals in IT management anizations utilize IT resourcesmore effectively. We also contribute to the literature by biningresearch from two distant literatures (IS and political sciences) [15],an approach that, to the best of our knowledge, few IS studies haveattempted.This study also offers several important managerial implications forthe public anizations. Our ndings emphasize that electedlawmakers be a key part of IT governance. They need to control andsupervise bureaucracy in IT management, so that IT resources are de-ployed and used at the areas that are mostly needed from the strategicstandpoint in a timely manner. It is also their responsibility to legitimizethe CIO position and to put a right person in that position who is capableof fullling the roles charged by the legislatures.The remainder of this paper anized as follows. The next sectionpresents our literature review and hypotheses. Section 3 introduces ourempirical approach. The result is presented in Section 4. Section 5 con-cludes the paper with discussions, limitations, and future researchdirections.2. Literature review and hypotheses2.1. Prior studies on IT governance anizational performanceWeill and Ross [73] dene IT governance as “specifying the decisionrights and accountability framework to encourage desirable behavior inusing IT”(p. 2). Several prior studies have focused on identifying anappropriate framework of IT governance in anizations[43,71–73]. For example, Brown and Magill [9], Sambamurthy andZmud [68], and Xue et al. [77] identify the antecedents for IT governancearrangement choices among various IT governance archetypes. Suchantecedents include corporate governance models, corporate strategies,IT knowledge of business divisions, and power of an anization. Xueet al. [78] study the relationship between environmental uncertaintyand IT governance centrality.Several studies in IT governance examine the impact of ITgovernance on rm performance. Gu et al. [27] demonstrate that ITgovernance misalignment, a discrepancy between an appropriateIT governance scheme and an actual governance mode, neutralizesthe contribution of IT to rm performance. Preston et al. [62] showthat a CIO's structural power within anization and a strongpartnership between the CIO and top management team (TMT) posi-tively contribute to the CIO's strategic decision making authority,which in turn positively affects the perceived contribution of IT torm performance. Banker et al. [3] reveal that alignment of a CIO–CEOreporting structure with a rm's strategic position has a positive impacton rm performance. Specically, they nd that a CIO–CEO reportingrelationship leads to higher rm performance under a differentiationstrategy, while a CIO–CFO relationship is positively associated withrm performance under a cost leadership strategy.Our study differs from these previous studies on IT governance intwo ways. First, we study the impact of IT governance on performancein the public sector, which is a new setting in IT governance research.We explain why cost efciency is a valid indicator for government per-formance in Section 2.2. Second, while most prior work in IT governancefocuses on the agency side of IT governance as in the power of CIO or ITmanagement centrality, we take a closer look at the principal side of ITgovernance with a lens of the principal–agent theory and investigatethe performance effect of legislative controls on IT management and in-vestments. In Section 2.3, we elaborate how the principals in govern-ments can supervise IT management.2.2. Cost efciency as government performanceState governments have been experiencing scal crises since theearly 1990s. By 1991, the total decits in the U.S. state and local govern-ments reached at $43.1 billion and have continued to rise since then[60,65]. This is attributed to, among others, growing senior population,mounting healthcare costs, and rising crime rates, all of which havecaused expenditure growth in such areas as public welfare and correc-tions [60]. Furthermore, Reschovsky [65] reports that state tax revenueshad declined by 20% from 2001 to 2003, while the real U.S. GDP had in-creased by about 15% for the same period. The expansion in federallymandated programs that are not fully subsided by the federal govern-ment has exacerbated this problem [2].In addition, in most U.S. states, state laws enact a variety of measuresto limit expenditure growth such as balanced budget requirements anddebt limitations [14,33]. For example, the Idaho State Constitutionrequires that its legislature pass a balanced budget. Louisiana statelaws mandate that “if a decit exists in any fund at the end of the scalyear, that decit shall be eliminated no later than the end of the next s-cal year.” According to the National Association of State Budget Ofcers,governors in 44 states must submit a balanced budget [47].Therefore, these factors – accumulated budget decits, declining taxrevenues, increasing demands for public services, and balanced budgetrequirements – pose elected executives, public ofcials, and politiciansin U.S. states an enormous challenge to improve cost efciency in stateadministration. In other words, the states have to provide a range ofpublic services that various laws and policies mandate to deliver withas limited resources as possible. This motivation to minimize costs isin contrast to for-prot rms, whose primary interests are to maximizesales revenues or prots, both of which are not a primary concern of thepublic anizations.In addition, while it may be possible to measure outputs or qualityof individual government services with such indicators as crime rate(public safety) or dropout rate (education), to the best of our knowl-edge, the literature in public economics or public administration doesnot offer a single quantitative measure, similar to sales or value-added,that accounts for diverse state government outputs such as transporta-tion or public welfare, each of which has widely different characteristicsfrom each other. Given our interest to examine the impact of IT invest-ments and governance on overall government performance, we posit275M.-S. Pang / Decision Support Systems 59 (–285here that cost efciency is a key performance measure that stategovernments aim to improve.The literature and anecdotal evidence point out that IT can be aneffective means for governments to improve cost efciency by automat-ing and streamlining manual, laborious business processes. Pandey andBretchneider [58] and Moon and Bretchneider [46] show that inefcien-cy is a primary motivation for governments to adopt and invest more inIT. For example, in 2005, the State of Arizona developed the Arizona 2-1-1 Online () to prehensive, statewide in-formation on health services and emergency operations to state citizens[50], so that the citizens can easily access the information and deal withurgent matters. The state reports that the immediate return on thisinvestment is substantial. The development and ve-year maintenancecosts are approximately $1 million [52], while annual savings in operat-ing costs of the 911 centers exceed $1.37 million [52]. Many stategovernments widely utilize decision support systems (DSS) equippedwith business intelligence capabilities in public welfare programs suchas Medicaid or food banks [51,54]. Such DSS based on data warehousesand analytic capabilities are used to streamline
applicationprocesses, to determine the eligibility of
provision in a speedymanner, and to detect fraudulent
claims. It is reported thatthese systems help the states save a substantial amount of operatingcosts in salaries and expenses that the states otherwise would incurwith outdated, paper-based processes.2.3. Bureaucracy in IT managementThe studies on IT management report that bureaucracy in ani-zations does exist in many forms. For example, IT managers may stick toan obsolete, inexible IT infrastructure that is costly to maintain. Theymay have vested relationships with certain vendors and offer them aspecial treatment in IT procurement processes. It may take severalmonths or years for them to make a critical decision in IT investmentand deployment because of anizational structures plex reporting processes. petent, underperforming IT staffmay not be disciplined or replaced at the right time [29]. IT managersmay refuse or fail to modate evolving needs of business managersand customers. Mark and Rau [36] state that “Business units can efrustrated by long delays in the deployment of needed capabilities, andIT may be viewed as an unresponsive bureaucracy, a black hole for busi-ness requests”(p. 23). Bureaucracy may lead to a failure in managing alarge-scale system development project, resulting in escalated develop-ment costs and long delays [55].Bureaucracy in IT management may take place in the business sideof anization as well. For example, uncoordinated IT investmentsand deployment are made by business units without close alignmentwith enterprise architectures or long-term strategies. Business man-agers may spend IT budgets in duplicate or patible systems,which do not contribute anization-wide synergies [69]. A partner-ship between IT and business functions, which is intended to curbbureaucracy in the IT function, can be a different manifestation ofbureaucracy [16]. Lohmeyer et al. [34] state that “Partnership betweenIT and business … can e plicated when -mittees proliferate and lengthy business plans pany each ITrequest. …With decision making so fragmented, few concrete decisionswere reached and nger-pointing was rampant”(p. 41). Our premise inthis study is that involvement and supervision of political principals inIT governance is an effective way to restrain such IT managementbureaucracy in the public anizations.To the best of our knowledge, there has been little research in ITmanagement bureaucracy in the public anizations. However,anecdotal evidence suggests that it does exist in government agenciesas well. For example, it has been reported that the recent failure in therollout of the U.S. federal healthcare exchange (healthcare.gov) wasattributed to poor project management, last-minute changes in require-ments, and failure in anticipating the number of users in the site [70].The integration project between the systems in the U.S. Department ofDefense and Department of Veterans' Affairs for seamless sharing ofdisabled veterans' information had experienced a long delay and con-siderable budget overrun, because of failure in upgrading the outdatedsystems and insufcient coordination between the two departments[44]. Similar IT management bureaucracy was reported in stategovernments as well [25].2.4. HypothesesThe theory of political controls on bureaucracy asserts that it is aresponsibility of elected representatives to curb bureaucracy and to con-trol agencies' decision makings and activities, so that policies and publicservices fulll the needs and desires of citizens, not of bureaucrats[11,38,45,75]. Specically, by bureaucracy, it refers to a moral-hazardsituation in which unelected bureaucrats (agents) seek to pursue theirown self-interests that conict with preferences of politicians andtheir constituencies (principals). This occurs because of informationasymmetry, where the principals do not have perfect informationon the agents' behavior and decisions, since the principals do not pos-plete knowledge and expertise in policy implementation.According to this literature, there are three major mechanisms withwhich politicians control bureaucrats —(i) monitoring and oversight,(ii) statutory controls, and (iii) approval of nomination of agencyexecutives.The legislature can exercise controls by directly monitoring andoverseeing executive agencies on an ongoing basis. Usually, monitoringand oversight are conducted by a mittee focusing on spe-cic policy and service areas such as education or transportation. mittee can carry out such activities as performance evaluation, bud-get analysis and approval, formal hearings, issuing policy guidelines,and carrying out a sanction or a punishment in case of a misconductor violation [40]. These supervising activities aim to make sure thatthe decisions and activities of executive agencies are congruent withthe goals and priorities put forth by the legislature and to detect andcorrect any pliance from the legislative purposes and guidelines.As of 2004, legislatures in 37 states have IT-mittees thatoversee state IT management [49]. The state legislature records revealthat such IT-related mittees can be an effectivemechanism to exert controls over IT management and operations. Forexample, the Kansas Legislature mittee on InformationTechnology, which is intended to review IT strategic plans, IT budgets,and the state enterprise IT architecture,mended in 2008 thestandardization of technology use and existing standards in Kansas hos-pitals (Page 4-1). It also requested that “all agencies will be expected toparticipate in the new statewide Financial Management System (FMS).Without full participation, multiple agencies will continue to incur sub-stantial, unnecessary costs”(Page 4-1). This demonstrates that theKansas Legislature requests utilization of IT to improve the quality andefciency of public health and nancial management and demandsthe establishment of a standard in technology use municationacross the state agencies to prevent wasteful spending in patible,silo systems. The Oregon Statutes Chapter 171 stipulates that the dutiesof Joint mittee on Information Management and Tech-nology include establishing “statewide goals and policy regarding” ITand making “mendations regarding established or proposed in-formation resource management programs and information technologyacquisitions.”2It further mandates that the enterprise management in IT“implement a state government-wide approach for managing distribut-ed information technology assets to minimize total ownership costs …while realizing maximum s for transacting the states businessand delivering services to its citizens.”31http://skyways.lib.ks.us/ksleg/KLRD/mRepts/jcit-cr.pdf (accessed on May. 9,.855 (accessed on Jun. 25, .477 (accessed on Jun. 25,
M.-S. Pang / Decision Support Systems 59 (–285We argue that efciency returns from IT spending are greater instates whose legislatures have an IT-specic mitteethan in others. It is not necessarily the case that state legislatures with-out an mittee do not play a controlling role for state IT functions.However, the political control theory points out that oversight andmonitoring are a challenging task for politicians. It is particularly thecase in IT management, where information asymmetry betweenpoliticians and managers is more severe. Thus, if state IT managementis overseen by a non-IT-mittee such as government opera-tion or economic development, mittee members, with limitedtime, information, and expertise, would not be able to exert as muchcontrol over IT management as they would be if they are in mitteeexclusively devoted to IT management. This leads us to propose thefollowing hypothesis.Hypothesis 1. All others being equal, the association between IT spendingand cost efciency is stronger in states whose legislature has an IT-mittee than in ones without such mittee.Another instrument for political controls is a statutory control [6] oran administrative procedure [38], in which politicians use legislation tostipulate rules and procedures that public ofcials are mandated tofollow in policy making and implementation. In effect, it takes a formof behavioral control on bureaucrats [20]. As mentioned above, directoversight and monitoring is expensive and imperfect for elected politi-cians because of information asymmetry and their limited resources.Thus, instead of costly intervention, the legislature uses statutorycontrols to dictate with what specic steps the agencies must take infullling legislative goals and directions. Bawn [6] explains that statuto-ry controls specify “how the agency anized, how the agency willmake the policy choice, who participates in agency decisions, whatinformation is used, and what qualications are required for keypersonnel”(p. 62).We argue here that establishing the position and duties of a CIO is aform of statutory control against a chief executive (governor) and seniormanagers in executive agencies and propose that efciency returns to ITspending are greater when the position and duties of CIO are formallyestablished by legislation than otherwise. By codifying the role andresponsibilities of CIO for statewide IT management and investments,the legislature constrains the discretion of other non-IT agency execu-tives, who might otherwise procure and deploy IT resources in anuncoordinated manner without taking an enterprise perspective and along-term strategic goal into account.As of 2004, 34 states establish the CIO positions by legislation, whilenine use an executive order to formalize the position. For example, theIowa Code Chapter 8A species the duties of the Director of InformationTechnology Services including “prescribe and adopt information tech-nology standards and rules” and “develop and mend legislativeproposals deemed necessary for the continued efciency of the depart-ment in performing information technology functions.” The MinnesotaStatutes Chapter 16E charges the state CIO to approve IT developmentprojects in state agencies and to develop cost-effective IT infrastructureand services to be shared across the state agencies. The CIO of NorthCarolina is also empowered by the General Statutes Chapter 147 tosuspend any IT project in executive agencies that does not conform tothe quality standard and enterprise architecture.All of this legislation above codies that the state CIO take greaterresponsibilities in statewide IT management vis-à-vis other seniorexecutives. This enables the CIO to control IT resources across anization and to formulate IT-related policies and principles forthe entire business functions to follow. The CIO will be a leaderand a facilitator in the efforts to achieve statewide efciencyimprovement with IT resources. With support and legitimacy con-ferred by legislation, he or she will be able to effectively coordinatestatewide IT operations and facilitate a standardized IT architecture.This will lead to greater anizational synergies [69] from ITresources across business units in multi-anizationssuch as state governments.The CIO position and duties can be instituted by an executive orderas well. For instance, the Michigan Executive Order 2001-3 creates theDepartment of Information Technology, which is headed by the stateCIO. Mayer [37] denes an executive order as “a presidential directivethat requires or authorizes some action within the executive branch”(p. 445). It has been used by the U.S. presidents to establish policy andto set up and alter administrative and regulatory processes [37]. Likepresidents, governors in U.S. states also have an authority to issueexecutive orders and generally are not required to seek an action orapproval of state legislatures in doing so. heless, even thoughexecutive orders carry an equal legal weight with legislation,4theirimpact and authority is more transient and provisional than legislation.An executive order can be revoked by eeding governors withoutlegislative consent, and Mayer [37] demonstrates that they are issuedmore frequently in the absence of public support.To be in charge of statewide IT management, the state CIO needssufcient authority and power granted by the legislative branch. It is aprerequisite to e resistance from business units which hadhad discretion in managing their own IT resources and investments.Legislative establishment of the CIO position, a type of statutory controlon agency executives, offers such authority and power to the CIO. Thus,we hypothesize the moderating role of legislative establishment of CIOposition as follows.Hypothesis 2. All others being equal, the association between IT spendingand cost efciency is stronger in states whose CIO position is established bylegislation than otherwise.A legislative branch controls bureaucracy by appointing agencyexecutives [11,75]. Although it does not directly appoint an executive,it may have an authority to approve appointment made by a chief exec-utive. This approval authority of the legislature discourages the chief ex-ecutive from nominating an appointee who is not likely to be approvedby the legislature. In this way, the elected lawmakers are able to choosea nominee who is expected to be more capable of fullling their policypreferences, and more importantly, they can select those who are notlikely to pursue their self-interests. Wood and Waterman's [75] empir-ical investigation shows that appointment approval is a more effectiveinstrument of political control than others such as direct oversight andstatutory controls.Based on the principal–agent model, Calvert et al. [11] set up a nom-ination game between a chief executive and a legislature, the latter ofwhich has a veto power for an appointee. According to their model, inequilibrium, there exists a pool of appointees whose preferences arenot too far from the preference of the legislature. It is mutually bene-cial for both the chief executive and the legislature to appoint andapprove one from the pool, respectively. In the absence of the vetopower, the chief executive would appoint a nominee whose preferenceis closer to his than to the legislature's. This model suggests that if thelegislature is to approve a CIO nominee, he or she is likely to better rep-resent its interests in regard to the state's strategic goals and directionsthan the one who is unilaterally appointed by the governor. Meanwhile,the appointing authority functions as a latent control as well [11], inwhich mid-level agency managers, who themselves are in the pool ofperspective appointees, voluntarily avoid to deviate from the prefer-ences of the legislature in their actions and decision makings.As of 2004, appointment of a state CIO requires legislative approvalin 19 states. As the political control theory predicts, state CIOs and ITmanagers are more likely to pursue the goal of legislatures in efciencyimprovement in IT management without bureaucracy when the CIOappointment is approved by state legislatures than otherwise. Thisleads us to offer the following hypothesis.4stion/040.html (accessed on Jul. 11, M.-S. Pang / Decision Support Systems 59 (–285Hypothesis 3. All others being equal, the association between IT spendingand cost efciency is stronger in states whose legislature approves a stateCIO nominee than otherwise.3. Empirical methodology3.1. The two-stage estimation approachThe estimation approach is based on Pang [59]. We estimate the re-lationship between IT investments and cost efciency and the moderat-ing effect of IT governance with a two-stage estimation approach basedon a multi-product translog cost function [12,13]. Several studies inpublic economics have used this method to measure governmentefciency [18,23,76]. For example, Geys [23] measures the cost efcien-cy of 304 Flemish local governments with a cost function model.Worthington [76] also adopts a similar two-stage approach to explainthe cause of inefciency in Australian local governments.The two-stage approach proceeds as follows. In the rst stage, we es-timate cost efciency of each state-year observation with a stochasticfrontier model [1,40]. In the second stage, we regress the estimated costefciency on IT spending and governance measures and control variables.Our key interest in measurement is technical cost efciency. Formal-ly, Koopmans [30] denes a producer as being technically efcient if andonly if it cannot increase production of an output without increasing anyinput or decreasing any other outputs. Kumbhakar and Lovell [32] de-ne a cost frontier as the least amount of inputs that can produce thegiven amount of outputs. Technical cost inefciency is measured asthe ratio of actual cost to the cost frontier. We estimate this technicalcost inefciency with a stochastic frontier model in the rst stage.A multi-product translog cost function with n outputs and m inputprices is given bylnCk;t
α0 Xni1αi lnYi;k;t 12Xni1Xnj1αij lnYi;k;t lnYk;t Xmi1βk;t12Xmi1Xmj1βk;k;t Xmi1Xnj1γk;t lnYk;t
εk;t1where k and t are subscripts for state and year, respectively. Ck,t is thetotal cost that state k incurs at year t, Yi,k,t are the amount of outputs,and wi,k,t are the input prices. Following Caves et al. [12], we impose con-straints for homogeneity of degree one in price on Eq. (1).5The use of a cost function in modeling state government productionis warranted as it implicitly assumes that outputs are exogenouslygiven. This is the case in our context because state government produc-tion is regulated by various federal and state laws, and it generally takesconsiderable time to amend such laws. Thus, compared to private-sector rms, governments cannot easily adjust the amount of outputsin response to changes in demands or input prices. In this regard, it isappropriate to assume that outputs (Yi,k,t) are exogenous, as in Eq. (1).The stochastic frontier model assumes a frontier to be stochastic, aseven the maximum production level may be inuenced by variousunobserved factors, random shocks, or statistical noises. Following thisrationale, Aigner et al. [1] and Meeusen and van den Broeck [41] assumethat a residual εk,t in Eq. (1) is given by εk,t = vk,t + uk,t. Here, vk,t repre-sents a random error and is assumed to follow a normal distribution ofN(0,σv2). uk,t refers to an inefciency factor, which by denition is greaterthan or equal to zero, and is assumed to follow a half-normal distribu-tion truncated below zero. The coefcients in Eq. (1) and the standarddeviation of the two error terms (vk,t and uk,t) are estimated usingmaximum likelihood estimation.6Technical cost inefciency (the ratioof the actual cost to the cost frontier) is given by exp(uk,t). We use anunbiased estimator for exp(uk,t) proposed by Battese and Coelli [4] toobtain technical inefciency Ineffk,t, of each state-year observation.For ease of interpretation, we reverse Ineffk,t by taking Effk,t =2
Ineffk,t as the dependent variable for the second stage estimation.This estimated technical efciency (Effk,t) is regressed on IT spending(IT), IT governance (g), and control variables (z) as shown in Eq. (2).In addition, the interaction terms of IT spending and governance vari-ables are included to test our hypotheses.Et
δIT ITk;t2Xδk;t2 XδITk;t2
ITk;t2 Xδk;t
ξk;t:2Following the notion in the IT value literature that it takes time for ITvalue to materialize [10], we use two-year lagged IT spending (ITk,t
2)and governance (gi,k,t
2) variables.7Since the hypotheses are testedwith a cross-sectional panel dataset and there may be unobservedheterogeneity in state government production, we estimate Eq. (2)with a xed-effects estimation model with Driscoll and Kraay standarderrors [19], which account for interstate correlation in residuals.3.2. Measures and data sourcesOur rst stage estimation with the cost function (Eq. (1)) adopts twoinput measures (m = 2) and four output measures (n = 4). First, atotal cost (C) is measured by the sum of per capita operation expensesand capital depreciation (buildings and equipments). Operationexpenses are obtained from the U.S. Annual State Government FinancesTable 1Data sources.Source Data VariableNational Association of State pendium of Digital Governments in the States () IT budget (IT1 and IT2) and IT governanceU.S. Census Bureau State Government Finances Operation expense (C), Capital price (w2), Federal grant (z4)State Government Employment & Payroll Labor price (w1)State Annual Population Estimate Population (z1)State Household e Household e (z2)State governments' Web prehensive Annual Financial Reports (CAFR) Capital depreciation (C)Alternative capital price (w2)Bureau of Economic Accounts State Gross Domestic Product (GDP) GDP (z3)Price Indexes for GDPNational Conference of State Legislature State legislature and gubernatorial election results Governor (z5), Legislature (z6)State Higher Education Executive Ofcers State Higher Education Finance Survey Education (Y1)Centers for Medicare & Medicaid Services National Health Expenditure Data by State of Resident Public Welfare (Y2)Federal Highway Administration Annual Highway Statistics Transportation (Y3)Bureau of Justice Statistics National Prisoner Statistics Public Safety (Y4)5This ensures that when all input prices wi are multiplied by x, the total cost C ismultiplied by x as well.6More details in the estimation process are available in Aigner et al. [1].7We re-estimated Eq. (2) with different lag lengths (from 0 to 4 years) and did notobtain qualitatively different results.278 M.-S. Pang / Decision Support Systems 59 (–285reports published by the U.S. Census Bureau. Capital depreciationgures are acquired from prehensive Annual Financial Reportsposted at states' Web sites (Table 1). Capital depreciation is reportedfrom scal-year 2001 to 2009, and some states do not post all nine-year reports at their Web sites, limiting our sample size to 428 state-years (Table 2). All dollar terms are adjusted for 2005 dollar.For output measures, we choose the four most representative publicservices that state governments supply — education, public welfare,transportation, and public safety.8Following prior studies in public eco-nomics, we selected four proxy variables to measure the four outputs asshown in Table 3. We also selected two input price measures in the costfunction estimation — capital and labor (Table 3). The correlationsamong the variables in the rst-stage estimation are presented inTable 4.The measures in the second-stage estimation (Eq. (2)) are shown inTable 5. IT spending, our key independent variable, is measured in twoways — per capita IT budget of central IT ofces (IT1) and the ratio ofIT budget to total general expenditures (IT2). These gures were obtain-ed from the pendium of Digital Governments in States pub-lished in 2003 and 2005 [49]. This publication reports IT budgets(central IT ofces and executive branches) in more than 40 statesfrom the scal year 2001 to 2005. However, there are many missingvalues in executive branch IT budget data.9Thus, we use the IT budgetof central IT ofces for our IT budget measures. The IT budget guresare available in 193 state-years between 2001 and 2005, biningthem with the sample in the rst-stage estimation left us 188 state-yearobservations. In addition, we had to drop three observations ofDelaware, which reports an unusually large amount of per capita ITspending (greater than 6σ). Hence, our sample size in the secondstage es 185. t-tests indicate that with respect to population,GDP, and total state expenditures, the states in our sample do not differsignicantly from those that are not in the sample.Data on the three IT governance variables – Committee (H1), Estab-lish (H2), and Appoint (H3) – were also obtained from pendium(Table 5). Committee is a dummy variable whose value is one if there isan IT-specic mittee in either state senate, house of rep-resentatives, or both. Establish is measured by one if a CIO position isestablished by legislation. It is also equal to one in ve states wherethe CIO position is established by both legislation and an executiveorder. Approve is equal to one if a state senate approves the nominationof a state CIO. Hypotheses 1, 2, and 3 will be supported if the coefcientsof IT × Committee, IT × Establish, and IT × Approve are positive andsignicant, respectively.The public economics literature provides economic, sociological, andpolitical factors that affect technical efciency in government produc-tion. Davis and Hayes [17] and Grossman et al. [26] suggest that thesize of jurisdiction (population) affects government efciency. Geys[23] and De Borger and Kerstens [18] argue that per capita elevel is related to efciency as well. Following these studies, we includepopulation, median household e, and per capita state GDP as con-trol variables in the second-stage estimation (Eq. (2)). The scal illusionhypothesis [23,26] suggests that a large inux of external revenues froma higher level of governments is a source of inefciency. Hence, we alsocontrol for per capita inter-government grants from the federal govern-ment to each state in Eq. (2). We also include Garand's [22] political in-dicators – a governor's party afliation and party control of legislatures– because they represent important political and institutional factorsthat may affect state government efciency. Eq. (2) also includes otherIT governance indicators – centrality of IT management, reporting rela-tionship of a CIO, and the size of central IT functions – as control vari-ables. Lastly, we include year dummies in our second-stage estimationto account for nation-wide changes in economic and political trends.Tables 5 and 6 show the summary statistics and correlations in thesecond-stage estimation, respectively.4. ResultsTable 7 shows the result of the rst-stage stochastic frontier estima-tion. Here, the variance of inefciency term (u) is positive and statisti-cally signicant, indicating possible presence of inefciency in stategovernment production. The estimated cost inefciency (the ratio of ac-tual costs to the cost frontier) ranges between 1.0165 and 1.9844.10Thisindicates that the most efcient and inefcient states spend 1.65% and98.44% more in operational expenses and capital depreciation than theminimum possible costs, respectively.The second-stage estimation results are presented in Table 8. InColumns (2) and (5), the coefcients of IT spending (IT1 and IT2) arepositive and statistically signicant. This shows that the more a stategovernment spends in IT, the more efcient it es. In bothColumns (3) and (6), we add the three interaction terms. Thecoefcients of IT1 × Establish and IT1 × Approve (Column 3) are posi-tive and signicant at the 1%-level of signicance, supportingHypotheses 2 and 3, respectively. This is the case with IT2 × Establishand IT2 × Approve (Column 6), showing that this result is robust tothe use of different denominators for normalization. It suggests thatwhen a state legislature establishes a CIO position by legislation orapproves a CIO appointee, the impact of IT budget on cost es stronger.Surprisingly, the coefcients of IT1 × Committee and IT2 ×Committee (Columns 3 and 6, respectively) are negative and signicant,Table 2States in the second-stage estimation.Geographic region and division are from the 2000 U.S. Census.Region Division StatesNortheast (1) New England Maine(4), New Hampshire(5), Vermont(3), Massachusetts(5), Rhode Island(5), Connecticut(3)(2) Mid-Atlantic New York(5), Pennsylvania(2), New Jersey(3)Midwest (3) East North Central Wisconsin(4), Michigan(5), Indiana(3), Ohio(5)(4) West North Central Missouri(5), North Dakota(5), South Dakota(5), Kansas(5), Minnesota(5), Iowa(5)South (5) South Atlantic Maryland(5), Virginia(3), West Virginia(2), North Carolina(5), South Carolina(3), ia(4), Florida(2)(6) East South Central Kentucky(5), Tennessee(5), Mississippi(5), Alabama(5)(7) West South Central Oklahoma(2), Texas(5), Arkansas(5)West (8) Mountain Idaho(5), Montana(5), Wyoming(3), Nevada(5), Utah(3), Arizona(5), New Mexico(5)(9) Pacic Washington(5), Oregon(3), California(3), Hawaii(5)The number in parentheses next to a state is the number of years that the state appears in the second-stage estimation.ording to the U.S. Census Bureau, the four service areas occupy as much as 65% ofthe total state general expenditures in the scal year 2008.9Executive branch IT budgets are available only for 29 states. Measuring IT spendingwith the total IT budget in both a central anization and executive branches leavesus only 92 observations in the second stage estimation, which are too few observationsto obtain statistically signicant estimations.10Eff (the dependent variable in the second stage estimation) is obtained byEff = 2
Inefciency. Thus, as in Table 5, Eff ranges from 0..9844) to0..M.-S. Pang / Decision Support Systems 59 (–285rejecting Hypothesis 1. This means that when state legislatures have anIT-mittee, efciency returns from IT investments diminish.One possible explanation for this is that the IT-specic -mittee may create another layer of bureaucracy in IT governance. Theactions and processes of mittee such as legislative hearingsand formal evaluation may hinder effective, timely IT investment anddeployment. Elected representatives in such mittee may not pos-sess sufcient knowledge and information to supervise state ITmanagement.In order to better explain this counterintuitive nding, we add thethree-way interaction term of IT × Committee × Approve to themodel. As shown in Columns (4) and (7) of Table 8, the coefcients ofthis three-way interaction term are found to be positive and signi-cant.11In both Columns (4) and (7), the coefcients of IT1 × Approveand IT2 × Approve e insignicant. This suggests that the positivemoderating effect of legislative approval for CIO (H3) disappears if thelegislature does not have a mittee devoted to IT manage-ment (Committee = 0). Likewise, in Column (4), the coefcient ofIT1 × Committee is 0.8–0.0033) if Approve is equal to 1,but it es 0.0033 if Approve is 0. This is the case in Column (7)as well. Hence, the presence of an IT-specic mitteestrengthens the relationship between IT spending and cost efciencyonly if the lawmakers approve a CIO appointment.To illustrate the effect of IT spending and IT governance on costefciency, we calculated an average cost reduction from a $1 in-crease in IT budget as follows. First, the cost frontier (the minimumpossible costs) of each observation is calculated by dividing the ac-tual costs by the estimated technical inefciency. For instance, if anactual per capita cost is $3000 and the estimated cost inefciency(the ratio of actual cost to the cost frontier) is 1.5, the cost frontierfor this state is $3000 / 1.5 = $2000. By averaging this cost frontierfor all state-year observations, we obtained $2545.33. When all ofApprove, Committee, and Establish are equal to 1, the sum of thecoefcients of IT1 and all interactions terms from Table 8, Column(4) is (0.0001 + 0.8
0.0033 + 0.0048) = 0.0036,meaning that a one dollar increase in per capita IT spending reducescost inefciency by 0.0036. Thus, in a state with the average cost frontier($2545.33), reduction in per capita cost from a $1 increase in per capita ITspending amounts to $2545.33 × 0.0036 = $9.16, as shown in Fig. 1-(a).Fig. 1-(b) is obtained from Table 8, Column (7) in a similar manner.Fig. 1-(a) and (b) illustrate the average cost reduction in each casewith 95% condence intervals when Establish is equal to one. Thisshows that average cost savings are higher when both Approve mittee are equal to one (Point [11]) than when only one of thesetwo governance controls is zero (Point [01] or [10]). It is interestingto see that in Fig. 1-(a), the expected cost savings from IT whenApprove = Committee = 1 (Point [11], $9.16) are similar to the caseof Approve = Committee = 0 (Point [00], $9.93). In Fig. 1-(b), theexpected cost savings with Approve = Committee = 1 (Point [11],$8.28) are higher than with Approve = Committee = 1 (Point [00],$7.03). However, in both Fig. 1-(a) and (b), the 95% condence internalsare narrower at Point [11] than at Point [00].We have several explanations for this nding. First, as we mentionabove, parison between Points [00] and [01] in Fig. 1-(a) and(b), the presence of a mittee on IT management plexity in IT management. Second, parison be-tween Points [00] and [10], it appears that the approval of CIOappointment by elected lawmakers weaken returns from IT spendingslightly, as shown by the negative but statistically insignicant coef-cients of IT × Approve in Columns (4) and (7). It may be becausemost elected representatives do not possess sufcient expertise in ITmanagement. A way to e these adverse effects is for themboth to engage in a CIO selection and to oversee IT management via mittee at the same time. For approval of CIO nominationto be effective, there must be a group of lawmakers dedicated to over-seeing IT management. Lastly, it can be shown that when Establish is0, the expected cost savings are less than $1, regardless of the valuesof Approve mittee, suggesting that formalizing the CIO positionby legislation is a key prerequisite to the positive impact of IT spendingon state cost efciency.5. Discussions and conclusionIt has been reported that due to severe scal crises for the last severalyears, IT investments in many governments became an early target ofbudgetary cuts in an attempt to close the budget gaps. For example,the NASCIO 2010 State CIO Survey reports that 64% of the state CIOs ex-pect reduction in IT budget [53]. In 2011, the State of Washington cutthe size of personnel in the Department of Information Services bymore than 10% [24]. Given this scal environment, elected ofcialsand public sector managers will ask how declining IT budgets should11We also tried to add other three-way interaction terms (IT × Establish × Approve andIT × Committee × Establish), which were estimated to be insignicant.Table 3Variable denition and summary statistics for the rst-stage estimation.Variables N Avg. Std. dev. Min. Max.Cost (C) 428 6.556 2.813The sum of per capita annual current operation expense and capital depreciation (building and equipments)Education (Y1) 428 6.81 50.29The number of enrolled students in public postsecondary educational institutions per thousand populationPublic Welfare (Y2) 428 141.87 55.16The number of Medicaid recipients per million population [18]Transportation (Y3) 428 5.729 266..43The length (mile) of state-maintained highways and roads per million population [76]Public safety (Y4) 428 1.575 281.9The number of inmates in state correctional facilities per million populationLabor price (w1) 428 .2 5231.4The monthly total payroll ($) divided by the number of fulltime-equivalent employees [17,76]Capital price (w2) 428 4...3The annual interest payments divided by mean debt level (average of beginning-of-scal-year debt and end-of-scal-year debt) [17,76]Fiscal year ; annual capital depreciation (part of C) is missing at 22 state-year observations.Table 4Correlation table for the rst-stage estimation.C Y1 Y2 Y3 Y4 w2Y1 0..6Y3 0.2 0..2 0.7w1 0.6 0.8 0..1 0.7 0.8280 M.-S. Pang / Decision Support Systems 59 (–285be managed and used more judiciously for the governments' attempt e the scal stress. This paper seeks to provide an answer to thisquestion.We draw upon the theory of political control on bureaucracy toarticulate the role of elected lawmakers in IT governance and to hy-pothesize the moderating effects of three legislative controls to ITmanagement on the cost-efciency returns from IT expenditures.Our theoretical reasoning proposes that by exerting controls overbureaucracy in IT management, politicians can ensure that IT man-agement and investments plish the required purposes moreeffectively, one of which is to provide public services for citizenswith as limited resources as possible, i.e. enhancing cost efciency.Our empirical analyses demonstrate that IT spending in U.S. stategovernments positively affects cost efciency in state governments.We further nd that this relationship between IT and cost efciency isgreater when the CIO position and duties are formalized by legislation.Indeed, among our three indicators for IT governance, the legislativeestablishment demonstrates the strongest moderating effect (Table 8).Furthermore, we nd that supervision of state legislatures on IT man-agement and approval of a CIO plement each other. Theimpact of IT expenditures on state efciency is greater when the statelegislature both has an IT-specic mittee and approvesnomination of a state CIO than when it plays only one of the two roles(Fig. 1).Our study carries some limitations. First, it bears the weaknesses ofstochastic frontier estimation and two-stage estimation approach[32,35,57], such as inconsistency of technical inefciency estimators orpossible biases from functional misspecication. However, despite ings, we believe that this approach is one of the most feasibleways to estimate the relationship between IT investments and govern-ment performance. Second, due to data limitation, we could use only ITbudget data of central anizations, as the pendium doesnot report the entire IT budget across state agencies and departments inmany states. But we still believe that our IT budget measure reects theTable 6Correlation table for the second-stage estimation.Eff z1 z2 z3 z4 z5 z6 z7 z8 z9 IT1 IT2 g1 g2z1 0..5z3 0.1 0..9 0.4z5 0.3 0.6 0..9 0.1 0.2z7 0.8 0.7 0.5 0..0 0.3 0.9 0.8z9 0.5 0.8 0.7 0.1 0..5 0.7 0.0 0.2 0.7IT2 0.5 0.2 0.4 0.1 0.4 0..9 0.1 0.9 0.0 0.5 0.4g2 0.8 0.0 0.4 0.5 0.1 0.4 0..1 0.3 0.4 0.5 0.4 0.3 0.4Table 5Variable denition and summary statistics for the second-stage estimation.Variables N Avg. Std. dev. Min. Max.Technical efciency (Eff) 185 0.7 0.52
(the ratio of actual cost to the cost in the frontier)Population (z1) 185 5.2 0.3Annual state population estimate (in millions)Household e (z2) 185 46.5 32.7State median household e (in thousand dollar)GDP (z3) 185 39.5 26.3Per capita state annual gross domestic product (in thousand dollar)Federal grant (z4) 185 1.6 0.3Per capita annual intergovernmental revenues from the federal government (in thousand dollar)Governor (z5) 185 0.7 0 11 if governor is Republican, 0 otherwiseLegislature (z6) 185 0.5 0.3The sum of the proportion of Republican lawmakers in state senate and that of state house of representativesCentrality (z7) 185 5.4 0 13The number of IT management areas that a state CIO is directly in charge of statewideReporting (z8) 185 0.5 0 11 if the state CIO directly reports to the governor, 0 otherwiseITEmp (z9) 185 0.1 0.1The ratio of central anizational personnel to total state personnel (%)IT1 185 19.4 0.5Per capita central IT ofce budget (in dollar)IT2 185 0.4 0.8The ratio of a central IT ofce budget to total general expenditure (%)Committee (g1) 185 0.3 0 11 if either state senate, house of representative or both has an IT-specic mittee, 0 otherwiseEstablish (g2) 185 0.4 0 11 if a state CIO position is established by legislation, 0 otherwiseApprove (g3) 185 0.2 0 11 if a state senate approves nomination of a state CIO, 0 otherwiseFiscal year
with a two-year lag of IT budget and governance ().281M.-S. Pang / Decision Support Systems 59 (–285extent to which state governments perceive IT as strategic resources forstatewide performance improvement. Lastly, we cannot rule out the pos-sibility that IT spending improves state cost efciency promisingthe quality of public services, although we do not expect that this is thecase. Further studies are warranted to examine whether government ITspending leads to an improvement in public service quality.This study contributes to the literature on IT business value and ITgovernance as follows. First, we expand the scope of IT value researchinto the public sector domain. Although the public anizationshave adopted IT in almost every aspect of administration as for-protrms do, the literature does not illuminate whether IT spending doescontribute to performance improvement in government. Second, weinvestigate how involvement of a legislative body in IT governance af-fects returns to IT spending. To the best of knowledge, the studies in ITgovernance to date have paid less attention to the role of “principals”–board of directors in for-prot rms or legislatures in governments –whose primary responsibility is to make sure that the activities anddecisions of managers are in accordance with the principals' interests.For this purpose, we bring the political control theory to theorize therole of the principals in IT governance and to hypothesize its moderat-ing effect. We believe that ours is the rst to adopt this interdisciplinaryapproach marrying the IS and political sciences literature.This paper also offers several meaningful implications for managersin the public anizations. Whether in the public or the businesssector, there may exist IT an IT function could beunresponsive to strategic needs of business units in IT, fall behindemerging trends in technology, or fail to manage capable employeesand vendors. In a large, multi-anization such as a stategovernment, IT bureaucracy may hamper coordination and standardi-zation in IT infrastructures and platforms across business units, whichis instrumental for cross-unit synergies [69]. For instance, effective im-plementation of a large-scale information systems, such as decision sup-port systems for public welfare administration that consist of datawarehouses and business analytics tools (Section 2.1), requires state-wide data collection and integration across state executive agencies,strong leadership of a state CIO, and cooperation of state agencies, allof which can hardly be achievable in the presence of IT bureaucracy.Our ndings suggest that for bureaucracy in IT management to be e, the IT function be panied with adequate supervision andguidance from elected representatives. The politicians should play akey role in IT governance for achieving greater returns to IT by selectinga right person for a CIO position, by legitimizing the CIO position andduties, and by monitoring IT operation on an ongoing basis. The caveat,however, is that involvement of the lawmakers in IT management mayunintentionally create another layer of bureaucracy. Thus, the IT gover-nance needs to strike the right balance between effective oversight on ITmanagement and speed investments and deployment in state-of-the-art IT systems.We believe that there are numerous research opportunities in ITgovernance and management in the public anization. First,Table 7The rst-stage stochastic cost frontier estimation results.Stochastic frontier estimation (dependent variable: log C)Model w/o interaction terms Model with interaction terms(1) (2) (3)ln Y1 0.8)ln Y1 29.8)ln w1 1.5)ln Y2 0.6)ln Y2 6.2)ln w2 2.5)ln Y3 0.4)ln Y3 10.5)ln w1 ln w1 0.0)ln Y4 0.0)ln Y4 4.2)ln w2 ln w2 0.6)ln w1 0.1)ln Y1 ln Y1 0.7)ln w1 ln w2 0.7)ln w2 0.1)ln Y1 ln Y2 0.6)ln w1 ln Y1 2.1)Controls Year/geographic divisions ln Y1 ln Y3 0.6)ln w1 ln Y2 0.9)ln Y1 ln Y4 0.0)ln w1 ln Y3 0.3)ln Y2 ln Y2 0.7)ln w1 ln Y4 0.4)ln Y2 ln Y3 0.7)ln w2 ln Y1 0.9)ln Y2 ln Y4 0.8)ln w2 ln Y2 0.0)ln Y3 ln Y3 0.6)ln w2 ln Y3 0.1)ln Y3 ln Y4 0.4)ln w2 ln Y4 0.5)ln Y4 ln Y4 0.6)Controls Year/Geographic divisionsσva0.7)σva0.9)σub0.5)σub0.3)ln L 146.2038 ln L 286.8115Wald χ2 1891.57 Wald χ2 2545.13N = 428; Standard errors are in parentheses. p b 0.1. p b 0.05. p b 0.01.aThe variance of idiosyncratic errors (vk,t);bThe variance of technical inefciency terms (uk,t, signicance from a log-likelihood test).282 M.-S. Pang / Decision Support Systems 59 (–285future research may directly study bureaucracy in IT management.What's the nature of IT management bureaucracy and what causes it?We do not expect that bureaucracy in IT management exists only in gov- it may take place in for-prot rms as well. Future researchmay study how chief executives, business managers, and board of direc-tors or legislatures can address IT management bureaucracy. Second, re-searchers may study the relationship between the legislature and theCIO. Specically, it would be interesting to study how involvement ofpoliticians in IT management affects the power and authority of theCIO. Third, researchers may generalize our study by investigating howboard-of-directors in business rms, who play a role of principals aselected politicians do in governments, affect IT management and busi-ness value from IT investments [56,67].Lastly, future research may investigate the effect of IT spending andIT governance mechanisms on other performance indicators in the pub-lic sector. Governments pursue far more diverse goals than for-prot or-ganizations. In addition to efciency, they aim at providing qualitypublic services, promoting economic development, and preservingequity, fairness, and rule of law. Interesting research questions to IS re-searchers include what impact IT spending and IT governance togetherhave on effectiveness in administration and public service delivery oroverall social welfare. For example, it would be interesting to see howIT investments improve the satisfaction of Medicaid beneciaries oraffect the effective operation of the Health Insurance Exchange that,mandated by the Affordable Care Act, state governments will be incharge of.Table 8The second-stage estimation results.Dependent variable — technical efciencyMethod Fixed effects regression with Driscoll–Kraay standard errors(1) (2) (3) (4) (5) (6) (7)Population 0.6)0.4)0.6)0.2)0.3)0.7)0.4)GDP 0.3)0.2)0.4)0.6)0.2)0.2)0.4)e 0.7)0.7)0.8)0.8)0.7)0.8)0.9)Federal grant 0.2)0.5)0.5)0.4)0.5)0.9)0.7)Governor 0.0)0.2)0.3)0.3)0.9)0.2)0.4)Legislature 0.4)0.5)0.8)0.9)0.5)0.2)0.7)Centrality 0.2)0.2)0.2)0.2)0.2)0.1)0.1)Reporting 0.9)0.6)0.9)0.9)0.7)0.1)0.6)ITEmp 0.6)0.5)0.5)0.5)0.5)0.6)0.6)Establish 0.3)0.2)0.5)0.1)0.2)0.5)0.7)Approve 0.6)0.4)0.5)0.4)0.6)0.7)0.0295(0.mittee 0.1)0.5)0.7)0.4)0.4)0.2)0.4)IT1a0.1)0.1)0.1)IT1 × Establish 0.0)0.9)IT1 × Approve 0.6)0.2)IT1 × Committee 0.8)0.8)IT1 × Approve × Committee 0.0)IT2b0.3)0.6)0.4)IT2 × Establish 0.1)0.2)IT2 × Approve 0.7)0.9)IT2 × Committee 0.9)0.2)IT2 × Approve × Committee 0.4)Controls Year Year Year Year Year Year YearF 68.56 70.98 47.88 53.19 70.02 52.42 51.74Within R20.5 0.3 0.8 0.4142N = 185; standard errors are in parentheses.Year dummies are omitted. AR(2) and spatial (interstate) correlation in residuals are assumed. p b 0.1. p b 0.05. p b 0.01.aPer capita budget of the central IT function.bThe ratio (%) of the central IT function budget to total expenditures.283M.-S. Pang / Decision Support Systems 59 (–285播放器加载中，请稍候...
该用户其他文档
下载所得到的文件列表IT governance and business value in the public anizations — The role of elected representatives in IT governance and its impact on IT value in U.S. state governments.pdf.pdf
文档介绍：
IT governance and business value in the public anizations — The role of elected representatives in IT governance and its impact on IT value in U.S. state governments.pdf IT governance and business value in the public anizations — The role of elected representatives in ITgovernance and its impact o...
内容来自淘豆网转载请标明出处.}