Empirical Issues on Capital Structure (1)nHow to measure capital structure?nDifferent definitionsnDifferent accounting practicesnWhich factors determine capital structure?nFirm-specific factorsnIndustry-specific factorsnCountry-specific (institutional) factors1Empirical Issues on Capital Structure (2)nWhy firms choose to issue a particular security like debt or equity?nWhat is the capital market reaction to the news of security offerings?nHow can we explain the price change?nIs the price change truly reflects future firm performance?2Why firms choose to issue debt or equity?nTrade-off hypothesisnBenefits (tax savings) and costs (bankruptcy costs, agency costs) of debt are “traded-off“ against one another.nBoth equity and debt issues are necessary to keep the firm on target.nPecking order hypothesisnManagers follow a financing hierarchy, and prefer to issue at first the security with the lowest information asymmetry.nDebt is preferred; Equity is issued as a last resort. nInformation asymmetry hypothesis nManagers are better informed than outside investors. Their choice of a financing source conveys new information.nSecurity issues result in outsiders adjusting their valuation.3Why firms choose to issue debt or equity?nOverinvestment (agency) hypothesisnFirms invest the proceeds from the issue in non-value maximizing projects.nEquity issue is valuable only when firm has good investment opportunitiesnWindow of opportunity (timing) hypothesisnFirms decide to issue a security when conditions are favourable.nIssue when a security is overpriced4Jung, Kim and Stulz, JFE, 1996.nDatanSample consists of 192 equity issues and 276 bond issues in the USnExclude utilities and banksnMethodologynCompute abnormal stock returns during and after the issue announcement using event study methodologynCalculate various firm characteristics of equity and bond issuing firmsnExamine the factors that can explain the issue of equity/ debt by performing logit regressions (models used to find which firms are likely to experience the event).5Kabir and Roosenboom, JCF, 2003.nDatanDutch listed companiesnSample consists of 58 equity (rights) issuesnExclude non-financial firmsnAnnouncement dates are collected from daily newspapersnStock prices collected from “Datastream”nFinancial statement data are collected from Yearbook of companies and Annual Reports6MethodologynCompute excess stock returns during and after the issue announcement using event study methodologynCollect information on different characteristics of equity issuing firmsnExamine by means of cross-sectional regression analysis the factors related to the announcement returnnEstimate excess operating performance of firms surrounding the time of equity issues 7Event StudynAlso known as Prediction error, Abnormal return or Excess return analysis. nIt is an inter-temporal research method.nIts purpose is to analyze the change of a variable (price/volume) over a period of time around the occurrence of an eventnAn event is a surprise (unexpected) happening that results in (new) information.8Steps in an Event StudynIdentify the event (announcement date)nModel the reactionnSelect the estimation and the event (test) periodsnEstimate abnormal (excess, predicted or residual) returnsnAnalyze the results9Methodological issues in Event StudynIdentification of the eventnSingle event / Type of eventnEasy / ProblematicnConfounding / ClusteringnStock return calculationnAdjust for capital changes (e.g. stock splits)nSimple (discrete) / Compound (continuous)nMeasurement intervalnMonthly / Weekly / Daily / Intra-day10nChoice of modelnMean Adjusted Model: ARi,t = Rit - RinMarket Adjusted Model: ARit = Rit - RmtnRisk Adjusted ModelnMarket Model: ARit = Rit - i - iRmtnCapital Asset Pricing Model: ARit = Rit - Rft - i(Rmt-Rft)nMulti-factor Model (Fama French Model): ARit = Rit - Rft - 1(Rmt-Rft) - 2RSMBt - 3RHMLtnMatched / Control Portfolio Model: ARit = Rit - RptMethodological issues (contd.)11nChoice of market indexnAggregating abnormal returnsnInfrequent (thin) / Non-synchronous tradingnDimson Aggregate Coefficient adjustmentnScholes-Williams AdjustmentnStatistical testsnParametricnt-testnNon-parametricnSign-testMethodological issues (contd.)12Testing statistical significancenPerforming a t-testi = 1, 2, ., N and t = 20, , +2013 Perform a sign-testZ = 2 | P - 0.5 | Np = proportion of positive excess returnsTesting statistical significance14Merit and deficiency of Event StudynMeritnInformation content testnMarket efficiency testnDeficiencynThe method can detect a change, but can not identify and isolate the cause of the change15Cross-sectional researchnEvent study helps us only to measure and test abnormal returns (AR). But, we also want to understand them as well.nFor example: why does one company have a big AR as a response to a news, whereas another company has a small AR as a response to the same news?nIn order to explain the sources of abnormal returns obtained from an e