1、本科毕业论文(设计)外文翻译原文THEDETERMINANTSOFCAPITALSTRUCTURECHOICETHISPAPERANALYZESTHEEXPLANATORYPOWEROFSOMEOFTHERECENTTHEORIESOFOPTIMALCAPITALSTRUCTURETHESTUDYEXTENDSEMPIRICALWORKONCAPITALSTRUCTURETHEORYINTHREEWAYSFIRST,ITEXAMINESAMUCHBROADERSETOFCAPITALSTRUCTURETHEORIES,MANYOFWHICHHAVENOTPREVIOUSLYBEENANALYZ
2、EDEMPIRICALLYSECOND,SINCETHETHEORIESHAVEDIFFERENTEMPIRICALIMPLICATIONSINREGARDTODIFFERENTTYPESOFDEBTINSTRUMENTS,THEAUTHORSANALYZEMEASURESOFSHORTTERM,LONGTERM,ANDCONVERTIBLEDEBTRATHERTHANANAGGREGATEMEASUREOFTOTALDEBTTHIRD,THESTUDYUSESAFACTORANALYTICTECHNIQUETHATMITIGATESTHEMEASUREMENTPROBLEMSENCOUNTE
3、REDWHENWORKINGWITHPROXYVARIABLESINRECENTYEARS,ANUMBEROFTHEORIESHAVEBEENPROPOSEDTOEXPLAINTHEVARIATIONINDEBTRATIOSACROSSFIRMSTHETHEORIESSUGGESTTHATFIRMSSELECTCAPITALSTRUCTURESDEPENDINGONATTRIBUTESTHATDETERMINETHEVARIOUSCOSTSANDBENEFITSASSOCIATEDWITHDEBTANDEQUITYFINANCINGEMPIRICALWORKINTHISAREAHASLAGGE
4、DBEHINDTHETHEORETICALRESEARCH,PERHAPSBECAUSETHERELEVANTFIRMATTRIBUTESAREEXPRESSEDINTERMSOFFAIRLYABSTRACTCONCEPTSTHATARENOTDIRECTLYOBSERVABLETHEBASICAPPROACHTAKENINPREVIOUSEMPIRICALWORKHASBEENTOESTIMATEREGRESSIONEQUATIONSWITHPROXIESFORTHEUNOBSERVABLETHEORETICALATTRIBUTESTHISAPPROACHHASANUMBEROFPROBLE
5、MSFIRST,THEREMAYBENOUNIQUEREPRESENTATIONOFTHEATTRIBUTESWEWISHTOMEASURETHEREAREOFTENMANYPOSSIBLEPROXIESFORAPARTICULARATTRIBUTE,ANDRESEARCHERS,LACKINGTHEORETICALGUIDELINES,MAYBETEMPTEDTOSELECTTHOSEVARIABLESTHATWORKBESTINTERMSOFSTATISTICALGOODNESSOFFITCRITERIA,THEREBYBIASINGTHEIRINTERPRETATIONOFTHESIGN
6、IFICANCELEVELSOFTHEIRTESTSSECOND,ITISOFTENDIFFICULTTOFINDMEASURESOFPARTICULARATTRIBUTESTHATAREUNRELATEDTOOTHERATTRIBUTESTHATAREOFINTERESTTHUS,SELECTEDPROXYVARIABLESMAYBEMEASURINGTHEEFFECTSOFSEVERALDIFFERENTATTRIBUTESTHIRD,SINCETHEOBSERVEDVARIABLESAREIMPERFECTREPRESENTATIONSOFTHEATTRIBUTESTHEYARESUPP
7、OSEDTOMEASURE,THEIRUSEINREGRESSIONANALYSISINTRODUCESANERRORSINVARIABLEPROBLEMFINALLY,MEASUREMENTERRORSINTHEPROXYVARIABLESMAYBECORRELATEDWITHMEASUREMENTERRORSINTHEDEPENDENTVARIABLES,CREATINGSPURIOUSCORRELATIONSEVENWHENTHEUNOBSERVEDATTRIBUTEBEINGMEASUREDISUNRELATEDTOTHEDEPENDENTVARIABLETHISSTUDYEXTEND
8、SEMPIRICALWORKONCAPITALSTRUCTURETHEORYINTHREEWAYSFIRST,ITEXTENDSTHERANGEOFTHEORETICALDETERMINANTSOFCAPITALSTRUCTUREBYEXAMININGSOMERECENTLYDEVELOPEDTHEORIESTHATHAVENOT,ASYET,BEENANALYZEDEMPIRICALLYSECOND,SINCESOMEOFTHESETHEORIESHAVEDIFFERENTEMPIRICALIMPLICATIONSWITHREGARDTODIFFERENTTYPESOFDEBTINSTRUM
9、ENTS,WEANALYZESEPARATEMEASURESOFSHORTTERM,LONGTERM,ANDCONVERTIBLEDEBTRATHERTHANANAGGREGATEMEASUREOFTOTALDEBTTHIRD,ATECHNIQUEISUSEDTHATEXPLICITLYRECOGNIZESANDMITIGATESTHEMEASUREMENTPROBLEMSDISCUSSEDABOVETHISTECHNIQUE,WHICHISANEXTENSIONOFTHEFACTORANALYTICAPPROACHTOMEASURINGUNOBSERVEDORLATENTVARIABLES,
10、ISKNOWNASLINEARSTRUCTURALMODELINGVERYBRIEFLY,THISMETHODASSUMESTHAT,ALTHOUGHTHERELEVANTATTRIBUTESARENOTDIRECTLYOBSERVABLE,WECANOBSERVEANUMBEROFINDICATORVARIABLESTHATARELINEARFUNCTIONSOFONEORMOREATTRIBUTESANDARANDOMERRORTERMTHEREIS,INTHISSPECIFICATION,ADIRECTANALOGYWITHTHERETURNGENERATINGPROCESSASSUME
11、DTOHOLDINTHEARBITRAGEPRICINGTHEORYWHILETHEIDENTIFYINGRESTRICTIONSIMPOSEDONOURMODELAREDIFFERENT,THETECHNIQUE“FORESTIMATINGITISVERYSIMILARTOTHEPROCEDUREUSEDBYROLLANDROSSTOTESTTHEAPTOURRESULTSSUGGESTTHATFIRMSWITHUNIQUEORSPECIALIZEDPRODUCTSHAVERELATIVELYLOWDEBTRATIOSUNIQUENESSISCATEGORIZEDBYTHEFIRMSEXPE
12、NDITURESONRESEARCHANDDEVELOPMENT,SELLINGEXPENSES,ANDTHERATEATWHICHEMPLOYEESVOLUNTARILYLEAVETHEIRJOBSWEALSOFINDTHATSMALLERFIRMSTENDTOUSESIGNIFICANTLYMORESHORTTERMDEBTTHANLARGERFIRMSOURMODELEXPLAINSVIRTUALLYNONEOFTHEVARIATIONINCONVERTIBLEDEBTRATIOSACROSSFIRMSANDFINDSNOEVIDENCETOSUPPORTTHEORETICALWORKT
13、HATPREDICTSTHATDEBTRATIOSARERELATEDTOAFIRMSEXPECTEDGROWTH,NONDEBTTAXSHIELDS,VOLATILITY,ORTHECOLLATERALVALUEOFITSASSETSWEDO,HOWEVER,FINDSOMESUPPORTFORTHEPROPOSITIONTHATPROFITABLEFIRMSHAVERELATIVELYLESSDEBTRELATIVETOTHEMARKETVALUEOFTHEIREQUITYINTHISSECTION,WEPRESENTABRIEFDISCUSSIONOFTHEATTRIBUTESTHATD
14、IFFERENTTHEORIESOFCAPITALSTRUCTURESUGGESTMAYAFFECTTHEFIRMSDEBTEQUITYCHOICETHESEATTRIBUTESAREDENOTEDASSETSTRUCTURE,NONDEBTTAXSHIELDS,GROWTH,UNIQUENESS,INDUSTRYCLASSIFICATION,SIZE,EARNINGSVOLATILITY,ANDPROFITABILITYTHEATTRIBUTES,THEIRRELATIONTOTHEOPTIMALCAPITALSTRUCTURECHOICE,ANDTHEIROBSERVABLEINDICAT
15、ORSAREDISCUSSEDBELOWSIXMEASURESOFFINANCIALLEVERAGEAREUSEDINTHISSTUDYTHEYARELONGTERM,SHORTTERM,ANDCONVERTIBLEDEBTDIVIDEDBYMARKETANDBYBOOKVALUESOFEQUITYALTHOUGHTHESEVARIABLESCOULDHAVEBEENCOMBINEDTOEXTRACTACOMMON“DEBTRATIO“ATTRIBUTE,WHICHCOULDINTURNBEREGRESSEDAGAINSTTHEINDEPENDENTATTRIBUTES,THEREISGOOD
16、REASONFORNOTDOINGTHISSOMEOFTHETHEORIESOFCAPITALSTRUCTUREHAVEDIFFERENTIMPLICATIONSFORTHEDIFFERENTTYPESOFDEBT,AND,FORTHEREASONSDISCUSSEDBELOW,THEPREDICTEDCOEFFICIENTSINTHESTRUCTURALMODELMAYDIFFERACCORDINGTOWHETHERDEBTRATIOSAREMEASUREDINTERMSOFBOOKORMARKETVALUESMOREOVER,MEASUREMENTERRORSINTHEDEPENDENTV
17、ARIABLESARESUBSUMEDINTHEDISTURBANCETERMANDDONOTBIASTHEREGRESSIONCOEFFICIENTSDATALIMITATIONSFORCEUSTOMEASUREDEBTINTERMSOFBOOKVALUESRATHERTHANMARKETVALUESITWOULD,PERHAPS,HAVEBEENBETTERIFMARKETVALUEDATAWEREAVAILABLEFORDEBTHOWEVER,BOWMANDEMONSTRATEDTHATTHECROSSSECTIONALCORRELATIONBETWEENTHEBOOKVALUEANDM
18、ARKETVALUEOFDEBTISVERYLARGE,SOTHEMISSPECIFICATIONDUETOUSINGBOOKVALUEMEASURESISPROBABLYFAIRLYSMALLFURTHERMORE,WEHAVENOREASONTOSUSPECTTHATTHECROSSSECTIONALDIFFERENCESBETWEENMARKETVALUESANDBOOKVALUESOFDEBTSHOULDBECORRELATEDWITHANYOFTHEDETERMINANTSOFCAPITALSTRUCTURESUGGESTEDBYTHEORY,SONOOBVIOUSBIASWILLR
19、ESULTBECAUSEOFTHISMISSPECIFICATIONTHEVARIABLESDISCUSSEDINTHEPREVIOUSSECTIONSWEREANALYZEDOVERTHE1974THROUGH1982TIMEPERIODTHESOURCEOFALLTHEDATAEXCEPTFORTHEQUITRATESISTHEANNUALCOMPUSTATINDUSTRIALFILESTHEQUITRATEDATAAREFROMTHEUSDEPARTMENTOFLABOR,BUREAUOFLABORSTATISTICS,“EMPLOYMENTANDEARNINGS“PUBLICATION
20、THESEDATAAREAVAILABLEONLYATTHEFOURDIGITSICCODEINDUSTRYLEVELFORMANUFACTURINGFIRMSFROMTHETOTALSAMPLE,WEDELETEDALLTHEOBSERVATIONSTHATDIDNOTHAVEACOMPLETERECORDONTHEVARIABLESINCLUDEDINOURANALYSISFURTHERMORE,SINCEMANYOFTHEINDICATORVARIABLESARESCALEDBYTOTALASSETSORAVERAGEOPERATINGINCOME,WEWEREFORCEDTODELET
21、EASMALLNUMBEROFOBSERVATIONSTHATINCLUDEDNEGATIVEVALUESFORONEOFTHESEVARIABLESTHESEREQUIREMENTSMAYBIASOURSAMPLETOWARDRELATIVELYLARGEFIRMSINTOTAL,469FIRMSWEREAVAILABLESECTIONDISCUSSEDANUMBEROFATTRIBUTESANDTHEIRINDICATORSTHATMAYINTHEORYAFFECTAFIRMSCAPITALSTRUCTURECHOICEUNFORTUNATELY,THETHEORIESDONOTSPECI
22、FYTHEFUNCTIONALFORMSDESCRIBINGHOWTHEATTRIBUTESRELATETOTHEINDICATORSANDTHEDEBTRATIOSTHESTATISTICALPROCEDURESUSEDTOESTIMATETHEMODELREQUIRETHATTHESERELATIONSBELINEARTHEMODELWEESTIMATEISANAPPLICATIONOFTHELISRELSYSTEMDEVELOPEDBYKJORESKOGANDDSORBOMITCANBECONVENIENTLYTHOUGHTOFASAFACTORANALYTICMODELCONSISTI
23、NGOFTWOPARTSAMEASUREMENTMODELANDASTRUCTURALMODELTHATAREESTIMATEDSIMULTANEOUSLYINTHEMEASUREMENTMODEL,UNOBSERVABLEFIRMSPECIFICATTRIBUTESAREMEASUREDBYRELATINGTHEMTOOBSERVABLEVARIABLES,EG,ACCOUNTINGDATAINTHESTRUCTURALMODEL,MEASUREDDEBTRATIOSARESPECIFIEDASFUNCTIONSOFTHEATTRIBUTESDEFINEDINTHEMEASUREMENTMO
24、DELTHEPARAMETERSOFOURMODELCANBEESTIMATEDBYFITTINGTHECOVARIANCEMATRIXOFOBSERVABLEVARIABLESIMPLIEDBYTHESPECIFICATIONOFTHEMODELTOTHECOVARIANCEMATRIXSOFTHESEVARIABLESOBSERVEDFROMTHESAMPLEINTHELISRELSYSTEM,THISISDONEBYMINIMIZINGTHEFUNCTION,FLOGDETTRS1LOGDETSPQ,WITHRESPECTTOTHEVECTOROFPARAMETERSOFTHEMATRI
25、CESREFERREDTOABOVETHISFITTINGFUNCTIONISDERIVEDFROMMAXIMUMLIKELIHOODPROCEDURESANDASSUMESTHATTHEOBSERVEDVARIABLESARECONDITIONALLYMULTINORMALLYDISTRIBUTEDOURESTIMATESOFTHEPARAMETERSOFTHEMEASUREMENTMODELAREPRESENTEDINTABLESIIANDIIITHEESTIMATESAREGENERALLYINACCORDWITHOURAPRIORIIDEASABOUTHOWWELLTHEINDICAT
26、ORVARIABLESMEASURETHEUNOBSERVEDATTRIBUTESBOTHTHEDIRECTIONANDTHEMAGNITUDE,ASWELLASTHESTATISTICALSIGNIFICANCE,OFTHEESTIMATESSUGGESTTHATTHESEINDICATORSCAPTURETHECONCEPTSWEWISHTOCONSIDERASDETERMINANTSOFCAPITALSTRUCTURECHOICETHEESTIMATESOFTHESTRUCTURALCOEFFICIENTSAREPRESENTEDINTABLEIVTHESECOEFFICIENTSSPE
27、CIFYTHEESTIMATEDIMPACTOFTHEUNOBSERVEDATTRIBUTESONTHEOBSERVEDDEBTRATIOSFORTHEMOSTPART,THECOEFFICIENTESTIMATESFORTHELONGTERMANDSHORTTERMDEBTRATIOSWEREOFTHEPREDICTEDSIGNHOWEVER,MANYOFTHEESTIMATEDCOEFFICIENTSAREFAIRLYSMALLINMAGNITUDEANDARESTATISTICALLYINSIGNIFICANTINPARTICULAR,THEATTRIBUTESREPRESENTINGN
28、ONDEBTTAXSHIELDS,ASSETSTRUCTURE,ANDVOLATILITYDONOTAPPEARTOBERELATEDTOTHEVARIOUSMEASURESOFLEVERAGEMOREOVER,THEESTIMATEDMODELSEXPLAINVIRTUALLYNONEOFTHECROSSSECTIONALVARIATIONINTHECONVERTIBLEDEBTRATIOSANEXAMINATIONOFTHECORRELATIONMATRIXOFTHESAMPLEDATATABLEVPROVIDESSOMEINSIGHTSABOUTTHEROBUSTNESSOFOURRES
29、ULTSPARTICULARLYNOTEWORTHYISTHEHIGHNEGATIVESIMPLECORRELATIONBETWEENOI/TAANDTHEVARIOUSDEBTRATIOSTHISRELATIONCANPOTENTIALLYCREATEAPROBLEMININTERPRETINGTHECORRELATIONBETWEENVARIABLESSCALEDBYEITHEROIORTAANDTHEDEBTRATIOMEASURESTHEBESTEXAMPLESOFTHISARETHEINDICATORSOFNONDEBTTAXSHIELDSFORINSTANCE,THESIMPLEC
30、ORRELATIONBETWEENNDT/TAANDTHEDIFFERENTMEASURESOFLEVERAGEISSTRONGLYNEGATIVEWHILETHISCORRELATIONISPREDICTEDBYTHEDEANGELOANDMASULISMODEL,ITSHOULDBENOTEDTHATTHELARGENEGATIVECORRELATIONMAYBEDUETOTHELARGEPOSITIVECORRELATIONBETWEENOI/TAANDNDT/TACAUSEDBYTHEIRCOMMONDENOMINATORSINTHEESTIMATEDSTRUCTURALMODEL,W
31、HEREWECONTROLFORTHEPROFITABILITYATTRIBUTETHATISMEASUREDBYOI/TAANDOI/S,THECOEFFICIENTESTIMATEFORTHENONDEBTTAXSHIELDATTRIBUTEISNOTSTATISTICALLYSIGNIFICANTMOREOVER,IFWEREPLACETHEDENOMINATORSOFTHENONDEBTTAXSHIELDINDICATORSWITHOI,THESIMPLECORRELATIONSARESTILLJUSTASSTRONGBUTAREREVERSEDFOREXAMPLE,NDT/OIISS
32、TRONGLYNEGATIVELYCORRELATEDWITHOI/TAANDSTRONGLYPOSITIVELYCORRELATEDWITHTHEMEASURESOFLEVERAGEUSINGINDICATORSSCALEDBYOIFORTHENONDEBTTAXSHIELDATTRIBUTELEADSTOPOSITIVECOEFFICIENTESTIMATESTHATARESOMETIMESMARGINALLYSTATISTICALLYSIGNIFICANTINTHESTRUCTURALEQUATIONSWHILETHISRESULTISINCONSISTENTWITHTHEDEANGEL
33、OANDMASULISMODEL,ITISMOSTLIKELYCAUSEDBYTHEWAYTHEVARIABLESUSEDASINDICATORSARESCALEDTHISPAPERINTRODUCEDAFACTORANALYTICTECHNIQUEFORESTIMATINGTHEIMPACTOFUNOBSERVABLEATTRIBUTESONTHECHOICEOFCORPORATEDEBTRATIOSWHILEOURRESULTSARENOTCONCLUSIVE,THEYSERVETODOCUMENTEMPIRICALREGULARITIESTHATARECONSISTENTWITHEXIS
34、TINGTHEORYINPARTICULAR,WEFINDTHATDEBTLEVELSARENEGATIVELYRELATEDTOTHE“UNIQUENESS“OFAFIRMSLINEOFBUSINESSTHISEVIDENCEISCONSISTENTWITHTHEIMPLICATIONSOFTITMANTHATFIRMSTHATCANPOTENTIALLYIMPOSEHIGHCOSTSONTHEIRCUSTOMERS,WORKERS,ANDSUPPLIERSINTHEEVENTOFLIQUIDATIONHAVELOWERDEBTRATIOSTHERESULTSALSOINDICATETHAT
35、TRANSACTIONCOSTSMAYBEANIMPORTANTDETERMINANTOFCAPITALSTRUCTURECHOICESHORTTERMDEBTRATIOSWERESHOWNTOBENEGATIVELYRELATEDTOFIRMSIZE,POSSIBLYREFLECTINGTHERELATIVELYHIGHTRANSACTIONCOSTSSMALLFIRMSFACEWHENISSUINGLONGTERMFINANCIALINSTRUMENTSSINCETRANSACTIONCOSTSAREGENERALLYASSUMEDTOBESMALLRELATIVETOOTHERDETER
36、MINANTSOFCAPITALSTRUCTURE,THEIRIMPORTANCEINTHISSTUDYSUGGESTSTHATTHEVARIOUSLEVERAGERELATEDCOSTSANDBENEFITSMAYNOTBEPARTICULARLYSIGNIFICANTINTHISSENSE,ALTHOUGHTHERESULTSSUGGESTTHATCAPITALSTRUCTURESARECHOSENSYSTEMATICALLY,THEYAREINLINEWITHMILLERSARGUMENTTHATTHECOSTSANDBENEFITSASSOCIATEDWITHTHISDECISIONA
37、RESMALLADDITIONALEVIDENCERELATINGTOTHEIMPORTANCEOFTRANSACTIONCOSTSISPROVIDEDBYTHENEGATIVERELATIONBETWEENMEASURESOFPASTPROFITABILITYANDCURRENTDEBTLEVELSSCALEDBYTHEMARKETVALUEOFEQUITYTHISEVIDENCEALSOSUPPORTSSOMEOFTHEIMPLICATIONSOFMYERSANDMAJLUFANDMYERSOURRESULTSDONOTPROVIDESUPPORTFORANEFFECTONDEBTRATI
38、OSARISINGFROMNONDEBTTAXSHIELDS,VOLATILITY,COLLATERALVALUE,ORFUTUREGROWTHHOWEVER,ITREMAINSANOPENQUESTIONWHETHEROURMEASUREMENTMODELDOESINDEEDCAPTURETHERELEVANTASPECTSOFTHEATTRIBUTESSUGGESTEDBYTHESETHEORIESONECOULDARGUETHATTHEPREDICTEDEFFECTSWERENOTUNCOVEREDBECAUSETHEINDICATORSUSEDINTHISSTUDYDONOTADEQU
39、ATELYREFLECTTHENATUREOFTHEATTRIBUTESSUGGESTEDBYTHEORYIFSTRONGERLINKAGESBETWEENOBSERVABLEINDICATORVARIABLESANDTHERELEVANTATTRIBUTESCANBEDEVELOPED,THENTHEMETHODSSUGGESTEDINTHISPAPERCANBEUSEDTOTESTMOREPRECISELYTHEEXTANTTHEORIESOFOPTIMALCAPITALSTRUCTURESOURCESHERIDANTITMANROBERTOWESSELS,1988“THEDETERMIN
40、ANTSOFCAPITALSTRUCTURECHOICE”THEJOURNALOFFINANCE,VOL43,NO1MAR,1988,PP119译文资本结构选择的决定因素本文分析了最近一些优化资本结构理论的解释能力。这项研究用三种方法扩展了对资本结构理论的实证研究。第一,它检验了更为广泛的资本结构理论集合,其中有许多还没有实证分析。第二,因为理论方面对于不同类型的债务工具有不同的经验影响,作者分析了短期、长期和可转换债券的措施,而不是债务总额措施。第三,研究采用因子分析技术,减轻了使用代理变数测量时遇到的问题。近年来,有一些理论被提出来解释整个企业的负债比率变化。这些理论认为,企业资本结构选择
41、的根据属性,决定了成本和与债务和股权融资相关的利益。在这方面的工作经验已经落后于理论研究,也许是因为公司相关属性都用相当抽象的概念来表示,而不是直接观察。其基本方法在以往的实证工作被用来与不可观察的理论属性代理估计回归方程,这个方法有一些问题。首先,可能没有我们想测量的具有独特代表性的属性。往往很可能代表了特殊属性,研究人员,缺乏理论的指导方针,可能受到诱惑而选择那些在统计拟合优度标准方面工作最好的变量,从而偏置了他们对他们测试显著性水平意义的解释。第二,往往很难找到与其他感兴趣的属性无关的特殊属性的措施。因此,选择代理变量可以测量多种不同属性的效果。第三,他们应该衡量的属性的不完全代表,他们
42、在回归分析中的应用中介绍了一个变量错误问题。最后,在代理变量中的测量误差可能与因变量的测量误差相关,建立假性相关与因变量无关,即使是在未观察的属性被测量到的时候。这项研究在三个方面扩充了对资本结构理论工作的经验。首先,它扩展了一些最近开发的理论研究到目前为止还没有实证分析的资本结构的决定因素的理论范围。第二,由于部分理论考虑到不同类型的债务工具的经验的影响,我们分析分成短期、长期和可转换债券不同的措施,而不是一个总债务总额的措施。第三,一种技术被明确承认并用来减轻上述讨论的测量问题。这种技术,是因子分析的方法测量到的或潜在变量的推广,被称为线性结构建模。很简单,此方法假设,虽然相关属性不能直接
43、观测,但我们可以观察大量一个或多个属性和随机误差项的线性函数的指标变量。还有就是,在本规格,与回报产生过程的直接类比假设持有套利定价理论。尽管我们的模型识别施加的限制是不同的,但评估技术与罗尔和罗斯用于测试的程序非常相似。我们的研究结果表明,具有独特或专门产品的企业具有相对较低的负债比率。唯一性是根据公司的研究和发展支出、销售费用以及员工自愿离职速度分类的。我们也发现,小公司比大公司更倾向于使用更多的短期负债。实际上我们的模型证明了在全公司可转换债券的比例没有变化,并认为没有任何证据支持理论工作,预测债务比率与一个公司的预期增长、非债务税盾、波动性或者其资产的抵押品价值相关。无论如何,我们这样
44、做,发现一些有利可图的企业的主张,即相对于他们的股票市值有较少的债务的支持。在本节中,我们提出了一个资本结构的不同理论认为可能会影响公司的债务权益选择属性的简短讨论。这些属性表示资产结构、非债务税盾、成长性、独特性、行业分类、规模、盈利波动性和盈利能力。这些属性,它们与最优资本结构的选择有关,他们的观察指标将在后面叙述。财务杠杆的六项措施将用于此项研究,它们是长期、短期以及可转换债券除以市价和权益帐面价值。尽管这些变量可以被合并提取一种共同的“负债比率“属性,这可能反过与对独立属性对立,我们完全有理由不这样做。资本结构的若干理论对不同类型的债务有不同的影响,而且,对于下面讨论的原因,结构模型的
45、预测系数可能会因为负债比率是在账面还是市场价值来衡量而有所不同。此外,因变量的测量误差均纳入扰动项,不偏向回归系数。测量数据的限制迫使我们依据账面价值测量债务,而不是依据市场价值。这也许就会得到更好的数据,如果市场价值对债务有效。然而,鲍曼表明账面价值与市场价值之间的相关性的研究非常多,所以由于帐面价值测度设定的误差是相当小的。此外,我们没有理由怀疑债务的市场价值和账面价值的研究差异与任何理论建议的资本结构的决定因素相关,所以没有明显的偏见,因为这将导致误设。1974年至1982年的这段时间,对前面的章节中讨论过的变量进行了分析。所有的数据除了辞工率是每年电子会计数据库工业文件。这些比率的数据
46、是来自美国劳动部门、劳动统计局,“就业与收入”出版社。这些数据仅在四位数(SIC代码)产业水平的制造业企业是可用的。从总样本中,我们删除了所有的观察结果,即在我们的分析中没有完整记录的变量。此外,由于大部分指标变量是按总资产或平均营业收入依比例决定的,我们被迫删除了少量的观察结果,其中包括这些变量中的一个负值小数目。这些要求可能对我们相对大的样本公司存有偏见,总的来说,469家公司是可行的。第二部分讨论了大量的属性和它们的指标,在理论上可能影响企业的资本结构的选择。不幸的是,理论不指定功能的形成,即描述属性如何与指标和债务率相关。用统计程序估计模型要求这些关系是线性的。我们估计的模型是约斯克格
47、和索伯开发的线性结构系统中的应用。它可以方便地认为是一个因素和分析两部分组成的模型测量模型和结构模型同时估计。在测量模型中,无法观察的企业特有属性用他们观察到的有关变量来衡量,例如,会计数据。在结构模型中,测量负债率被指定为在测量模型中定义的属性功能。我们的模型参数可以通过模型()暗示的可观察变量的协方差矩阵来估计,该模型规范了这些样本观察到的变量的协方差矩阵S。在线性结构统计系统中,这是通过减少功能完成的,FLOGDETTRS1LOGDETSPQ,关于上文提到的矩阵的参数向量。拟合函数是来源于最大似然程序,并假设可观察变量通常是有条件分布的。我们对模型参数的测量估计载于表二和三。这些估计通常
48、与我们先前关于指标变量测量未被观察到的属性多么好的想法一致。方向和规模以及统计意义,预算表明这些指标反映的概念,即我们要考虑的资本结构的选择决定因素。结构系数的预计是列于表四。这些系数详细说明了预计的所观察到的负债比率上未观察到的属性的影响。在大多数情况下,系数对长期和短期债务比率的估计是预测的标志。然而,许多估算系数数量相当小并且在统计上无关紧要。特别是,该属性代表非债务税盾,资产结构,并且它的变动显得与杠杆的各项措施无关。此外,预算模型说明该属性在可转换债券比率中几乎没有变化。一个样本数据(表五)的相关矩阵审查提供了一些关于我们研究结果稳健性的见解。尤其值得注意的是,OI/TA和各种负债比
49、率之间的高度相反单相关。这种关系有可能在解释变量由其他投资或技术援助决定和务比率衡量尺度之间相关关系时造成问题。最好的例子就是非债务税盾的指标。例如,NDT/TA和杠杆的不同措施之间的单相关关系呈强负相关性。尽管这种关联是由DEANGELO和MASULIS模型预测的,但应该指出的是强负相关性,可能是由于OI/TA和NDT/TA的公分母造成的正相关关系引起的。在估计的结构模型中,我们为盈利能力属性所控制的,是通过OI/TA和OI/S测量的,对非债务税盾属性的测量控制系数估计在统计上并不显著。此外,如果我们用OI取代非债务税盾指标的分母,简单的相关性依然强劲,但一样发生了逆转。例如,NDT/OI与OI/TA强负相关,而且与杠杆措施强正相关。为非债务税盾属性使用OI指示标尺,而导致积极的系数估计,有时在结构方程统计上显著。虽然这个结果与DEANGELO和MASULIS模型不一致,这很有可能是由按比例指标决定变量的方式引起的。本文介绍了估计不可观察的属性对企业的负债率选择影响因素的分析技术。虽然我们的研究结果不是决定性的,它们充当了证明经验规律性与现存理论相一致的文件。特别是,我们发现债务水平与公司业务线的“独特性”负相关。这方面的证据与特曼影响一致,企业可能会大量增加他们客户、员工和在清盘事件中负债率较低供应商的成本。研究结果还表明,交易成本可能是资本结构选择的