1、Socio-Economic Planning Sciences 37 (2003) 2943MeasuringDEAefciencyincabletelevisionnetworkfacilities:what are appropriate criteria for determining the amounts ofgovernmental subsidies?Toshiya Jitsuzumia,*, Akihiro NakamurabaFaculty of Economics, Nagasaki University, Katafuchi, Nagasaki 850-8506, Ja
2、panbInstitute for Posts and Telecommunications Policy, Azabudai, Tokyo 107-0052, JapanReceived 26 February 2001; received in revised form 8 July 2001AbstractRecenttechnologicaldevelopmentshavetransformedthecabletelevisionindustry(CATV)fromasimplere-transmitter of terrestrial broadcasting to a provid
3、er of a broader-band information infrastructure. Withcable operators facing an undesirable market situation while operating a ber optics network vital to thecreation of an information-based society within Japan, the Japanese Government has introduced severalsupportivemeasuresfortheindustry.Suchgover
4、nmentalinterventionshoulddesirablybejustiednotonlyfrom a political viewpoint but also from an economic one; otherwise such measures may distort economicefciency. In this article, using Data Envelopment Analysis (DEA), the authors identify the deciencies ofan accounting-based intervention scheme and
5、propose an alternative framework with no disincentive sideeffects.r 2002 Elsevier Science Ltd. All rights reserved.Keywords: Data Envelopment Analysis; Cable television; Yardstick competition1. IntroductionDuetorapidprogressinberopticanddigitalizationtechnologies,thecabletelevisionbusinesshas evolve
6、d from a simple re-transmitter of terrestrial broadcastingwhen introduced in Japanalmost a half-century ago, the industry sought mainly to improve household TV receptionto apromising provider of state-of-the-art broader-band interactive network facilities which supportvariousinformationtechnology(IT
7、)applications.Infact,226operatorshadbeengrantedlicensesfor telecommunications services as of the end of December 2000. These operators are currently*Corresponding author. Tel.:/fax: +81-95-820-6396.E-mail addresses: jitsuzumnet.nagasaki-u.ac.jp (T. Jitsuzumi), a-nakamurasoumu.go.jp (A. Nakamura).003
8、8-0121/02/$-see front matter r 2002 Elsevier Science Ltd. All rights reserved.PII: S0038-0121(02)00005-8providingorplantoprovideso-calledfullservice,includingbroadcasting,communications,and Internet connections. Their broader-band network is widely viewed as one of the mostimportant elements of soci
9、o-economic infrastructure in the Information Age. In addition,inasmuch as Nippon Telegraph and Telephone Corporation (NTT), the largest telecommunica-tions carrier in Japan, dominates the telecommunications market, the government expects theseinteractive CATV networks to apply competitive pressure t
10、o NTT and to fuel competitionin thismarket segment.Inresponsetothecurrentsituation,theJapanesegovernmenthasderegulatedtheexistinglegalframeworks to support the CATV industry. For example, the Ministry of Posts andTelecommunications (MPT)1has dramatically reduced market barriers for new entrants byes
11、tablishingastandardapplicationprocedureforpermissiontoinstallCATVfacilitiesin1994.Inthis regard, it published a Manual for Market Entry into the Cable TV Business in 1996, andsubsequently removed constraints on foreign investments in 1999.Nevertheless, many of Japans CATV businesses are currently su
12、ffering heavy nancial losses.Duetofactorssuchaslowmarketpenetration,2largedepreciationburdens,insufcientgrowthinviewing hours 1 and channel repertoire (i.e., the set of channels watched regularly by anindividual or household) 2, competition with terrestrial or satellite broadcasters, and generalecon
13、omic trends in Japan, only 63% of operators ended the scal year (1999) with a bottom-lineprot, and 3% with a cumulative surplus33. As this situation may jeopardize Japanscompetition and IT policies, and may further induce political concerns about uneveninfrastructuredevelopmentamonglocalcommunities,
14、theJapanesegovernmenthasimplementedseveral supportive measures that include subsidies and tax reductions.Such governmental measures should be preferably justied not only from a political viewpointbut also from an economic one. When verifying economic justications for given interventions,regulators n
15、eed to take the following steps:*First, to examine whether some form of market failure exists, such as economies of scale orexternalities; otherwise, the interventions will likely harm the Pareto optimum.*Next, to design an appropriate support mechanism that determines which rms are qualiedfor suppo
16、rt and in what amount, in order not to harm rms incentives for rationalization.In this article, we focus our efforts on the second area in seeking to identify an appropriatemechanismforoptimalintervention.Currently,mostgovernmentalsupportsseemtobedesignedto help CATV rms because they are nancially d
17、amaged. The government assumes that suchrmscannotaffordtoprovidesufcientnetworkcapacityforhouseholds(hereafter,referredtoasaccounting-based schemes). However, in the following sections, by showing that rms withnancial hardship are equally or more efcient in supplying network capacity than other rmst
18、hat are more protable, we show that an accounting-based mechanism will not motivate CATVrms to efciently provide network facilities. To overcome this deciency of accounting-based1In 2001, MPT was integrated with two other governmental bodies and reorganized into the Ministry of PublicManagement, Hom
19、e Affairs, Posts and Telecommunications.2As of March 2000, the number of subscriber households in Japan had reached 9.47 million (up 19.3% from theprevious year), yet market penetration remained in the 20% range.3The average rm suffered 21.4 billion yen of operating loss and 17.5 billion yen of ordi
20、nary loss, up 77% and 2%from the previous years gures, respectively.T. Jitsuzumi, A. Nakamura / Socio-Economic Planning Sciences 37 (2003) 294330schemes, we use Data Envelopment Analysis (DEA) and propose an alternative mechanism thathas no such shortcomings.DEA, rst developed by Charnes et al. 4, h
21、as been widely used to evaluate productionefciency/effectiveness of public sectors without assuming any functional specications betweeninputs and outputs. The decisional technique is now widely used for various Japaneseperformance analyses, as well. For example, Tone 5 has discussed various types of
22、 DEAapplications in both public and private sectors. Goto and Tsutsui 6 and Sueyoshi 7 haveutilized DEA to evaluate electric utility rms. In a series of studies, Sueyoshi 810 assessedJapanese telecommunications policy. While acknowledging his contribution to Japanesetelecommunications, we must menti
23、on that his empirical studies have paid relatively littleattention to the Japanese CATV industry. As far as we know, our research is the rst attempt toapply DEA to Japanese CATV rms. It is hoped that this effort will help guide governmentalpolicymaking in the Japanese CATV industry.Thearticleisorgan
24、izedasfollows:Section2introducesourapproachandtheDEAframeworkusedherein.InSection3,wecomparethenancialoutcomesofCATVrmstotheirDEAscores.Next, we present a proposal for an optimal support mechanism for network capacity provision.Results are summarized and future issues are briey discussed in the conc
25、luding section.2. Our approachBeforeexplainingouranalysisindetail,itisworthnotingtheuniqueelementsofaCATVrmsdecision-making process. This process consists of two parts, each having a completely differenttime perspective. One involves strategic decisions concerning the provision of network capacities
26、,and is extremely long-term in nature; the other consists of everyday managerial decisionsinvolving issues such as programming selection, marketing to potential viewers, and customersupport for subscribers. Given the rapid development of networking technology and the massivetime and investment neces
27、sary for network construction or renovation, it is rational for CATVrms to make strategic decisions based not on current market conditions but on expected futuredemand. Moreover, severe competition with terrestrial or satellite TVs compels CATV rms toquickly expand their network coverage regardless
28、of actual demand so as to be able to respondimmediately to potential customers. At the same time, however, day-to-day managerial decisionsmust be made while assuming that the capacity of network facilities is exogenously xed. Ourinterestsfocussolelyonthelonger(i.e.,strategic)perspectivebecauseithasd
29、irectimplicationsforthe evolution of domestic information infrastructure; however, a rms overall nancial positionderives from the combined outcomes of these two types of decisions.Our approach consists of two parts. First, we compared a rms nancial index with itstechnological efciency in network cap
30、acity provision (hereafter Network Capacity ProvisionEfciency Score, or NCAPES) and found little correlation between the two scores. As nancialguresdonotfunctionasgoodproxiesforNCAPESs,anaccounting-basedyardstickcompetitionmechanism cannot function well from the viewpoint of motivating CATV rms to e
31、fcientlysupply network capacity for households. Seeking to support network capacity provision withoutharmingefciency,westructuredanNCAPES-based mechanism,whichformsthesecondpartofour approach.T. Jitsuzumi, A. Nakamura / Socio-Economic Planning Sciences 37 (2003) 2943 31Theyardstickcompetition,whichi
32、sabasisofourproposal,wasoriginallyintroducedtoregulateregionalmonopolists.Itsessentialideaistocomparetheperformanceofoneregionalmonopolisticrm with that of others operating under similar conditions, in order to reduce informationasymmetries between the regulator and the regulated. Working from such
33、comparisons, theregulator can minimize the efciency deterioration inevitable in any monopolistic situation byawarding a bonus to better-performing rms. (For details, see Armstrong et al. 11)Although the yardstick competition was originally introduced to regulate lucrative regionalmonopolies, its bas
34、ic concept can also be applied to structure subsidies for unprofitable regionalmonopolists, such as those in Japans CATV industry. This concept suggests that, when anindustryhastobemaintainedforcertainreasonsandcurrentlyno(oronlyafew)rmsgenerateaprot, regulators should subsidize all rms within the i
35、ndustry by an amount equal to theaveraged loss of rms that achieve sufcient efciency. This is designed to motivate less efcientrms to follow best practices. When applying this concept, it is very important to nd anappropriate index that accurately reects certain aspects of the industry in question.
36、With theCATV industry, therefore, performance comparisons should be between NCAPES gures, notnancial outcomesunless the rms nancial gures are perfect proxies for NCAPES.Moreover, in the real world, as no two rms operate under the same exogenous conditions, aperformance index free of such exogeneity
37、must be prepared. As Japanese CATV operators aresubject to entry regulations, the exogeneity exists in area selection. It is thus not fair to includearea-related impacts in performance comparisons; if such factors are included, seriousdisincentives will likely affect environmentally advantaged rms.2
38、.1. The DEA modelIn order to measure NCAPESs, we relied on DEA. The basic concept here is to evaluate thecomparative efciency of a certain decision-making unit (DMU) by calculating a proportionbetween outputs and inputs. For a single-input, single-output case, the calculation is very simple.On the o
39、ther hand, with multiple inputs and/or outputs, one must create a virtual input and/oroutput using appropriate weights and divide the latter by the former; it is here that DEA istypically applied. Mathematically, DEA can be expressed as the linear programming problems,(1) and (2), with inputs, x1; x
40、2yxm; outputs, y1; y2yys; a non-negative vector l l1;y;lnT;and yC3as an optimal solution of the problem (1). Here, the superscript T indicates a vectortranspose. If an optimal solution satises yC3 1 and is zero-slack sC0C3 0;sC3 0; then theparticular DMU under study (DMUo) is called efcient. If DMUo
41、 is not efcient, severalefcient DMUs are designated as a reference set and a ynof DMUo is calculated based on adistance between DMUo and a production possibility frontier constructed by the reference set.minys:t: yx0C0XlX0;YlXy0;LpelpU;lX0 wheree 1;y;11maxl;sC0;s o esC0ess:t: sC0 yC3x0C0Xl;s YlC0y0l
42、X0;sC0X0;sX0:2T. Jitsuzumi, A. Nakamura / Socio-Economic Planning Sciences 37 (2003) 294332DEA has several variations. By setting L 0; U N; a CCR-I (input-oriented CharnesCooperRhodes 4) model assumes a constant returns to scale; meanwhile, when L U 1; aBCC-I(input-orientedBankerCharnesCooper12)mode
43、lassumesvariablereturnstoscale.Bymodifying the model in order to take uncontrollable variables into account, an input-orientednon-controllablevariablemodelunder aconstantreturns-to-scale assumption(NCN-I-C) model,or an input-oriented non-controllable variable model under a variable returns-to-scalea
44、ssumption (NCN-I-V) model, is obtained. Since the theoretical distribution of the efciencyscore in DEA is usually unknown, in order to compare efciencies between two types of DMUs,Brockett and Golany 13 proposed using the MannWhitney rank test that is nonparametric innature. (For further details, pl
45、ease refer to Cooper et al. 14)Considering a CATV rm as an infrastructure provider, we constructed the following models.(Forreference,we constructedmodelsinwhichaCATVrmis consideredasimplebroadcastingoperator; model frameworks and results are provided in Appendix A.)Infrastructure Model (I-Model):In
46、puts (5): Monetary value of broadcasting equipment.4Capacity of trunk network (MHz*km).Capacity of distribution lines (MHz*km).Number of staff members necessary for maintenance and technical support.Number of staff members necessary for program production and editing.Outputs (2): Total number of cha
47、nnels that subscribers can view (CH*households).Network capacity available for interactive services (MHz*households).5Reduced Infrastructure Model (RI-Model):Inputs (3): Monetary value of broadcasting equipment.Capacity of network (MHz*km).Number of staff members for maintenance and technical suppor
48、t, production,and editing.Outputs (2): Total number of channels that subscribers can view (CH*households).Network capacity available for interactive services (MHz*households).Itisworthstressingthattechnicalefciencyasmeasuredinthefollowingsectionsrelatestothelonger-term component of the CATV decision
49、-making process and is merely a prerequisite for arms overall nancial success. In order to improve nancial outcomes, a rms management hasto achieve sufcient overall efciency, which, within the DEA framework, can be divided intotechnical efciency and allocative efciency. The relationship between these efciency concepts is4Therelatedguresaresummedusingdepreciation,opportunitycostofcapital,leasefe
