1、会计实证研究基础框架和基础方法杨丹博士生导师、教授西南财经大学工商管理学院 执行院长财政部中国会计学术带头人中国金融会计学会专家委员会委员Email: 内容n会计实证研究理论框架n会计实证研究方法分类n会计实证研究的基本模型内容n会计实证研究理论框架n会计实证研究方法分类n会计实证研究的基本模型引言：资本市场实证研究重要性n实证研究的重要性n经验哲学和先验哲学n波普尔的证伪主义：经验证伪原则。他指出经验科学的特征在于它的方法,但这种方法并不是经验主义所主张的归纳法。波普尔认为从单称陈述中归纳出普遍陈述,这是不可能的。波普尔认为,科学的理论或命题不可能被经验证实,只能被经验证伪。没有提供证伪结构
2、的理论就是伪科学。研究划分n层次划分：基础研究、应用研究、开发研究、政策研究n领域划分：自然科学研究、社会科学、人文/艺术研究(一般不叫科学，但并非较低)、哲学研究n研究方法：运用科学方法的研究、运用其他方法的研究资本市场实证研究的重要性n资本市场提供了最综合性的对企业成功的衡量标准：n利润管理n现金管理n价值管理n上市公司上市公司 有形资产（亿元美金）有形资产（亿元美金） 帐面价值帐面价值 市场价值市场价值 美国在线美国在线 AOL 10 1200 可口可乐可口可乐 COCA-COLA 160 1600 迪斯尼迪斯尼 WALT DISNEY 150 420 福特汽车福特汽车 FORD MOT
3、OR 2820 430n管理思路即如何增加价值，局部的优化决策不一定增进价值。n会计信息观即考虑如何使会计信息有用。会计(社会责任)等信息和市场价值的关系就是资本市场实证研究的主要内容。-会计信息含量n投融资、股利和治理等对价值的影响就是财务管理（公司金融）实证研究价值创造过程的三大问题n资源问题：资本流动n动机问题：代理关系n经理人和股东：Jensen & Meckling 1976，-公司治理n股东和债权人：Smith & Warner 1979，-最佳合同n大股东和小股东：LLSV 2000，-投资者保护n信息问题：信息不对称（Akerlof 1970）-会计信息含量资本市场基本框架：P
4、aul M. Healy Krishna G. PalepuJournal of Accounting and Economics 31 (2001) 405440Research implicationsnThe information and agency frameworks raise a number of important questions for financial reporting and disclosure researchers. n(i) the role of disclosure and financial reporting regulation in mi
5、tigating information and agency problems,n (ii) the effectiveness of auditors and information intermediaries as a means of increasing the credibility of management disclosures and uncovering new information,n (iii) factors affecting decisions by managers on financial reporting and disclosure, andn (
6、iv) the economic consequences of 会计信息链条n信息传递者n信息编码n噪音n信息解码n信息使用者 会计理论的功能n指导（Guide）会计实务n评估（Evaluate）会计实务n解释（Explain）会计实务n预测（Predict）会计实务规范性（Normative/Prescriptive）理论实证性（Positive/Descriptive）理论会计理论的结构会计目的会计目的基本假设基本假设广泛原则广泛原则详细方法详细方法规范性规范性（演绎法演绎法）实证性实证性（归纳法归纳法）1、实证会计研究的演进n早期研究：收益决定模型（1920s -1930s），研究已经
7、存在的会计方法的理论原则，以及如何确定真实收益n决策有用性研究（1960年以后）：n（1）最初，描述现存的会计方法n（2）会计信息对经济决策的影响，即会计信息含量，主要基于CAPM和EMHn（3）解释会计管制：如政府对会计的管制及各种准则制定团体的出现n（4）会计对社会福利的影响，即会计准则的政治本质n（5）会计在契约中的应用（ Positive Accounting ）会计信息的功用：当代观点（1）信息观（Information perspective）对经济收益观批判的产物，信息观认为，在市场不完全和充满不确定性的现实世界中，任何会计方法都不可能得到企业的“真实收益”，会计信息的功用是向投
8、资者传递某种有助于判断和估计经济收益的“信号”而不是经济收益本身。投资者在接到该信号后，会产生信念的改变：a均值改变；b方差改变；c两者同时改变。实证模型：AR=a+bUE+代表人物：BEAVER等（2）计价观（valuation perspective）n历史上有两次：一次是古典计价观；一次是当代计价观。n区别：a、强调的内容不同； b、假设不同n c、实质不同：完全作用、作用程度n当代计价观与信息观的区别：前者指出会计盈余数字虽然不等于经济收益，但它赋予会计数字与股票价格之间的直接联系；后者会计盈余只是一种信号，这种信号必须通过某种更具资产价值属性的变量（股利）才能与股票价格相关联。n实证
9、方面：一般不做严格的区分，但是前者往往用股价作为因变量，而后者却用股价的变动作为因变量。换句话说，计价观研究中心在资产负债表；信息观则在损益表。nP=a+bE+ b的大小成为关注重点n代表人物：OHLSON、PENMAN等（3）契约观（contract perspective）n认为一切会计理论都是利益集团为推行符合自身利益的会计政策而制造的借口，因而是不科学的，只有研究利益集团之间对会计政策的主张和选择、旨在解释和预测会计实务而不是规范会计实务的会计理论才是客观和可行的。为此提出了positive accounting。n不仅分析了传统会计理论体系的缺陷，而且及时地在新的会计理论体系中引入企
10、业代理理论，以契约成本为核心理念，建立了实证会计理论框架。n企业是一组契约的集合，最难解决的是委托代理关系（逆向选择、道德风险）n契约理论是信息观和计价观的一个非常必要的补充。n代表人物：WATTS 、ZIMMERMAN会计信息：现金应计利润信息含量（增量信息）价格模型（存量信息）审计师：作为Y或X审计意见、审计收费、审计师选择概念：可靠、相关、一致、重要、可比、谨慎度量：资产、负债、盈余、应计利润、公允价值GAAP会计师：盈余管理和会计政策行为会计投资者：个人和机构行为研究公司投融资行为：公司治理盈余预测资本结构股利政策非会计信息：环保、社会责任有效率的经济会计信息加工使用过程的研究资本市场
11、实证研究n财务会计n审计n管理会计n财务管理（公司金融） 资本市场实证研究n财务会计 Acctg standards research Research on info. Content of listed companies Fin. analysts forecast research Acctg. info. disclosure research Decision-making usefulness researchn财务管理nMarket Efficiency nIPO and financing researchnCapital structure researchnCapital
12、 cost researchnInvestment strategy and corporate performance researchnCorporate governance researchnDividend policy research nRisk management管理会计n绩效评价n内部控制n成本会计n审计 Auditing quality research Auditors change research Auditing pricing research Auditing opining research Auditing function research (relat
13、ed to earnings management , corporate governance and the improvement of performance and info. quality, etc.) Info. contents of auditing report research Auditing standards researchn公司治理 Corporate governance with accountingCorporate governance with financeCorporate governance with auditing内容n会计实证研究理论框
14、架n会计实证研究方法分类n会计实证研究的基本模型会计实证研究方法分类规范研究与实证研究n规范研究(Normative Research)：世界应该怎样n实证研究(Positive Research)：世界是怎样n面对不同的情况，选择不同的研究范式n区分规范研究和实证研究的好坏，就象比较苹果和橘子哪个好吃一样Positive ResearchnAnalytical researchnModelingnDeductionnArchival researchnSurveys (e.g., mail questionnaire) nField testsnCase Study nExperiments
15、研究方法的三维度特性Scientist as ObserverNaturalistic ObservationLaboratory ResearchScientist as ParticipantTrue ExperimentField Research学术论文范式学术论文范式n1 1 演绎范式演绎范式：逻辑演绎或思辨的方法阐明思想（管理理论和管理哲学）n罗珉，刘永俊.企业动态能力理论架构与构成要素J.中国工业经济，2009年1月n2 2 数理范式数理范式：数学公式推导证明思想、数学要求相对较高（营销、物流、产业经济学）nYang Shilei, Charles L. Munson, Bint
16、ong Chen. Using MSRP to enhance the ability of rebates to control distribution channels. European Journal of Operational Research，205 (2010) 127135n3 档案类研究档案类研究：二手数据+计量方法或统计方法、计量要求相对较高（财务、金融、产业经济学 ）n杨丹，魏韫新，叶建明.股权分置对中国资本市场实证研究的影响及模型修正J.经济研究，2008年第三期n4 调研类调研类：问卷调查+统计方法或计量方法（营销、人力资源、财务、金融）n唐小飞，贾建民，周庭锐.
17、关系投资和价格促销的价值比较研究J.管理世界，2007年第五期n5 实验研究实验研究：实验室实验+统计方法或计量方法（营销、人力资源、财务、金融、经济学）n杜建刚,范秀成.服务消费中多次情绪感染对消费者负面情绪的动态影响机制J.心理学报,2009年第4期. n1、2一般称为理论研究；3、4和5一般称为经验研究Advantages of Experimental MethodnClear manipulation and control - make a strong case of causality and increase internal validity nE.g., studies
18、on Earnings ManagementnStudy the process of decision-makingnE.g., stock market reactions to accounting informationnAllows for “what if” scenarionE.g., study the effect of a newly proposed accounting standardChallenges in Using Experimental MethodnLimited External ValiditynDifficulty in designing exp
19、erimentsnIssues about participantsnIssues about materialsAdvantages of Survey MethodnRepresentative sample can be collected from the desired population in a variety of locations at low cost;nCan keep the richness of the real business world.Challenges in Using Survey MethodnLimited internal validity
20、nHard to conclude causalitynDemand or response biasnSelf-report biasnHypothesis guessing资本市场试验方法Vs档案式研究：优势n档案试验主要看X和Y是否关联，试验研究可以深入解决“为什么”的问题，提供解释的视角n试验研究可以进行过程研究，回答“How”的问题n在没有档案数据的时候，比如新准则刚使用、报告调整的时候仍可以研究n可以有效控制其他变量。试验方法Vs档案式研究：劣势n需要寻找样本n设计试验和执行成本高n风险较大n内部有效性和外部有效性的权衡，外部有效性较少。近期财务会计文章理论基础和研究方法近期财务会
21、计文章理论基础和研究方法Economic TheoryPsychology TheoryTotalArchival Method580 articles (98%)13 articles (2%)593 articlesExperimental Method 10 articles (29%)25 articles (71%)35 articlesTotal 590 articles38 articlesLisa Koonce & Molly Mercer (2004):Using Psychology Theories in Archival Financial Accounting Res
22、earch.内容n会计实证研究理论框架n会计实证研究方法分类n会计实证研究的基本模型基本模型一：ERC和实证会计的开始PerspectiveInformationMeasurementWhose point of viewuserspreparers and standard settersObjectiveuse accounting information to determine stock pricesuse stock prices as benchmark to determine the relevance of accounting informationResearch fo
23、cusmarkets response to announcement of accounting informationthe association between stock price (or return) and accounting informationTiming of informationImportantNot importantInformation vs. Measurement PerspectiveEarnings-returns relationship and the two perspectivesnReturn: measured by CAR, how
24、 long? nShort-window: earnings response coefficient (ERC) information perspectivenLong-window: earnings association coefficient measurement perspectivenIf measured as annual return change in price between two time points validation test in measurement perspectivenSometimes both are called (ERC)nERC
25、becomes a major part of information perspective researchHow to measure “unexpected earnings” (UE)nTime series modelnAnalyst forecastsnUE = E if E follows random walk48Evolution of ERC ResearchnBall and Brown (1968): + and -nBeaver, Clarke, and Wright (1979): rank correlation between R and E nBeaver,
26、 Lambert and Morse (1980): overall ERC (not firm-specific)nDeterminants of firm-specific ERCs: Kormendi and Lipe (1987) and Easton and Zmijewski (1989)nEarnings persistence nRisk (beta)nSizenCapital structurenetc.Why ERC is small? (in Kathoris survey)nPrices lead earnings (Beaver, Lambert and Morse,
27、 1980)nOnly part of earnings change is unexpectednDeficient GAAP (Lev 1989)nAccounting system too slow in reflecting economic realities nNoise in earningsnEarnings = “true earnings” + noisenQ: accounting system is counterproductive? nTransitory earningsnAsymmetry between positive and negative transi
28、tory earningsnMake earnings-return relationship nonlinear (Freeman and Tse 1992): high |UE| is less persistent than low |UE|, so ERC is “S” shape nMarket inefficientnProblem of cross-sectional modelsnExample: Teets and Wasley (JAE, 1996)Application of ERC in research (examples)nIncremental informati
29、on content of current cost earnings datanCorporate governance structure vs. information asymmetrynReporting of transitory gains and lossesnForeign GAAP, IAS, or US GAAP in cross-listing in USnCapitalization of R&D Ball and Brown (1968)nFirst empirical (“archival”) accounting researchnWhy it was revo
30、lutionary:In the 1960snNormative approach was hitting a dead-endnPessimistic view on the usefulness of accounting informationResearch QuestionsnDo stock prices respond to the release of earnings information?nDependent variable:nMarket-model abnormal returns cumulated by means of an Abnormal Performa
31、nce Index (API)nPresentation of results: graph of API over months -12 to +6 relative to the annual report announcement monthnIndependent Variable: Unexpected Earningsn3 measures of unexpected earnings:1) Market index model, based on NI: nFirm-specific time-series models, explaining 25% of the variat
32、ions (table 1)nBased on E (to avoid the autocorrelation in levels, table 2)2) Market index model, based on EPS 3) Random walk, based on EPS Derivation and Statement of Hypotheses nAny theoretical developments? Not possible at that timenBeavers and Ohlsons later efforts基本模型二：Ohlson模型The Feltham-Ohlso
33、n framework: Implications for empiricistsBernard, Victor LContemporary Accounting Research; Spring 1995; 11, 2; ABI/INFORM GlobalFeltham-Ohlson ModelingtititiEarningsAbnoamalFutureExpectedofPVBVV,结论：1.帐面价值（BV）与盈余均为重要定价因素。2.企业财务越健全，帐面价值的系数越低。3.股东权益报酬率低者（绩效较差者），盈余的系数较低。4.盈余预期为正时，其系数为正且显着，预期盈余为负时，其系数近于
34、零。5.盈余中现金流量及应计部分有重大的差别定价。Feltham-Ohlson Modeling杨丹、魏蕴新、叶建明（2009）n其中： MVEt代表t年公司实际市值nBVEt代表是t年公司账面价值nXt是t年公司净收益nDIVt是t年公司分配的股利nVt是t年公司收益不能观察到的部分01234tttttMVEBVEXDIVV基本模型三：盈余管理Earnings Management Research1.Positive Accounting Theory2.Healy and Wahlen (1999, AH)3.Do users see through earnings managemen
35、t?4.Methodology issues: Dechow, Sloan, and Sweeney (1995)(At the courtesy of Prof . Kevin Chen of HKUST)I. Positive Accounting Theory Ways of earnings management1. Accounting method choices or changes2. Real transactions (e.g., sale of assets, early retirement of debt)3. Accounting transactions (e.g
36、., restructuring, write-off)4. Accruals manipulation: current accruals (e.g., A/R) and noncurrent accruals (e.g., depreciation)Meaning of “earnings management research”nBroad and specific: from all above to just (4)n(1) is the subject of positive accounting theory (PAT)PAT and agency theory of the f
37、irmnCorporation is the combination of numerous contracts between different partiesnDebt agreements: contracts between corporation and lendernExecutive compensation plans: contracts between corporation and executivesnPolitical costs: “contracts” between corporation and governmentnAccounting numbers a
38、re used as the basis of the contractsnManagers (agents) have discretion in determining the accounting numbers Motivation of using the discretion:“opportunistic” vs. “efficient” behaviorn“Opportunistic behavior”: nmanagement can choose accounting methods for its own interest (assumed in PAT usually)
39、n “Efficient behavior”: nfor the benefit of the firm nSeparable sometimes, example: ncompany increases income to pay higher bonus and tax opportunistic nDifficult sometimes, example: nuse “operating leases” to reduce debt/assets ratio, efficient for shareholders and managers, but opportunistic to cr
40、editors.PAT researchnOriginal purpose of publishing JAEnDependent variable: naccounting choice or change (0/1)nThree main independent variables : ndebt agreements proxied by D/E ratio (to represent the “tightness” in lending agreement) nbonus plans observed from bonus payment nBecome a separate and
41、continuing line of researchnpolitical costs proxied by size (to represent the degree of political pressure).Common weaknesses of PAT researchnAssume opportunistic but ignore the efficient motivations.nDifficult to find precise proxy variables nExplain one accounting choice or change at a time replac
42、ed by earnings (accruals) managementRelation between PAT and “earnings management research”nPAT: starts with an accounting choice or change, and find out the motivation n“Earnings management”: starts with a motivation (e.g., reducing net income to get government relief) find out earnings manipulatio
43、n (usually from accruals).Abnormal accruals as measure of EM nJones (1991)nMotivations:nSample: 23 firms petitioned for import relief with International Trade Commission (ITC).nIn this investigation, firms have to show “injury” due to unfair foreign competition in order to get import relief.nITC doe
44、s not have motivation to look beyond earnings managementnTwo measures of EM:nChange in total accruals divided by total assets. nResults in Table 3: Decrease in year 0 (investigation year). Same pattern for earnings, cash flows, and sales.nModel of accruals (“Jones model”)nAssume that revenue and PPE
45、 are the drivers of accruals. All other change is due to manipulation.nResult: unexplained accruals are significantly negative in year 0.Executive compensation researchnSurviving line of research in PAT, why?nData availablenTheoretical basisnRelate to corporate governancenIssue: the role of accounti
46、ng income in the determination of compensationnLimitations:nY: stock options and share ownership dominate cash bonusnX: accounting income becomes less importantnMore in Bushman and Smith (JAE 2001)2. Healy and Wahlens review (1999, AH): four issues1. Incentives of earnings management (show the exist
47、ence of EM)2. Which accruals are managed?3. Frequency of earnings management4. Resource allocation consequences of earnings management 1. Incentives of earnings managementnCapital market motivationsnManagement buyouts: DeAngelo (1988), Perry and William (1994)nIPO, SEO: Rangan (1998), Teoh et al. (1
48、998 JF and JFE), Shivakumar (2000 JAE)nStock-financed acquisitions: Erickson and Wang (1998)nMeeting analyst forecasts: Burgstahler and Eames (1998)nContracting motivationsnCompensation: Healy (1985), Holthausen, Larcker, and Sloan (1995)nFirms close to dividend covenant: Healy and Palepu (1990)nFir
49、ms about to violate debt covenants: DeFond and Jiambalvo (1994)nCEO changes: Dechow and Sloan (1991)nRegulatory decisions nImport relief: Jones (1991)nCapital requirement in banks and insurance companies: manynAntitrust hearing : Key (1997) (cable TV industry)2. Which accruals are managed?nUsually a
50、bnormal accrualsnStart from Jones 1991nBad debt provision nMcNichols and Wilson (1988)nLoss reserves in banks and insurance companies: many studiesne.g., Collins, Shackelford and Wahlen (1995): no clear manipulation of loan loss reserves in banksne.g., Beaver and McNichols (1998): manipulation of cl
