Public traded companies are obliged to account for their operating numbers from time to time. These numbers are usually published in their interim and annual reports. Besides annual accounts and their balance sheet, the income statement provides great value for potential investors and shareholders. This master thesis wants to prove that announcements of operating numbers have verifiable influence on share prices. Event studies are mainly used to determine abnormalities in return series. An event study focuses on the prediction of normal returns with the help of a certain market model and ascertains abnormal returns in a second step. The selection of a suitable market model is the essence of every event study. On the one hand, there are market models which use certain external factors in their regression equation, having influence on returns. On the other hand, a widely range of autoregressive models computes returns on the basis of their own precursors. Furthermore an extension to that is even able to detect and map volatility clustering in return series. Eventually the variety of different market models exhibits that return prediction can only be an approach to real observations. Besides the study of abnormalities on a certain event day, it could be worthwhile to examine intervals in return series prior and afterwards an incident. Keynote of this analysis is that investors and shareholders could detect earnings surprises premature and also trade afterwards a publication on an extraordinary basis. The statistical question raised is whether there are coincidences between significantly more distinct trends in return series and the release of business reports. Furthermore, it is arguable whether these coincidences appear only on a random basis or not. In addition to that, time series of capital market values succumb specific statistical characteristics. Properties like a leptokurtic distribution and weak stationarity constitute prerequisites to subsequent analysis. Additionally autocorrelation of returns is taken into particularly consideration. To sum up, it seems that capital markets provide a diversity of attributes to analyse. Taken all together, these procedures try to disprove capital market efficiency.

Inhalt: 1 Einleitung 2 Überblick über den Stand der Forschung 3 Transmissionsmechanismus Vermögenspreise und Investitionen 3.1 Transmission über den Aktienmarkt 3.1.1 q-Kanal 3.1.2 Erwartungskanal 3.1.3 Bilanzkanal 3.2 Transmission über den Immobilienmarkt 3.2.1 Alternativer q-Kanal 3.2.2 Bilanzkanal 3.3 Gesamtwirtschaftliche Investitionsfunktion 4 Modellierung des vermögenspreisinduzierten Investitionseffektes 5 Ökonometrische Methodologie 5.1 Kointegration und Fehlerkorrekturmodell 5.2 Dynamisches OLS nach Stock und Watson 6 Statistische Datenbasis 7 Empirische Ergebnisse 7.1 Empirische Ergebnisse ausgewählter Studien 7.2 Empirische Ergebnisse für Deutschland 7.2.1 Test auf Integration 7.2.2 Investitionsmodelle 7.2.3 Ergebnisse DOLS 7.2.4 Test auf Parameterstabilität 7.2.5 Ergebnisse Impuls-Antwort-Analyse 7.3 Zusammenfassung der empirischen Ergebnisse 8 Geldpolitische Implikationen 9 Zusammenfassung

Public traded companies are obliged to account for their operating numbers from time to time. These numbers are usually published in their interim and annual reports. Besides annual accounts and their balance sheet, the income statement provides great value for potential investors and shareholders. This master thesis wants to prove that announcements of operating numbers have verifiable influence on share prices. Event studies are mainly used to determine abnormalities in return series. An event study focuses on the prediction of normal returns with the help of a certain market model and ascertains abnormal returns in a second step. The selection of a suitable market model is the essence of every event study. On the one hand, there are market models which use certain external factors in their regression equation, having influence on returns. On the other hand, a widely range of autoregressive models computes returns on the basis of their own precursors. Furthermore an extension to that is even able to detect and map volatility clustering in return series. Eventually the variety of different market models exhibits that return prediction can only be an approach to real observations. Besides the study of abnormalities on a certain event day, it could be worthwhile to examine intervals in return series prior and afterwards an incident. Keynote of this analysis is that investors and shareholders could detect earnings surprises premature and also trade afterwards a publication on an extraordinary basis. The statistical question raised is whether there are coincidences between significantly more distinct trends in return series and the release of business reports. Furthermore, it is arguable whether these coincidences appear only on a random basis or not. In addition to that, time series of capital market values succumb specific statistical characteristics. Properties like a leptokurtic distribution and weak stationarity constitute prerequisites to subsequent analysis. Additionally autocorrelation of returns is taken into particularly consideration. To sum up, it seems that capital markets provide a diversity of attributes to analyse. Taken all together, these procedures try to disprove capital market efficiency.:1 Einleitung 1 2 Theoretische Grundlagen 3 2.1 Bilanzanalyse 3 2.1.1 Rechte und Pflichten 3 2.1.2 Kennzahlen 4 2.2 Analyse des Kapitalmarktes 5 2.2.1 Aktienanalyse und Markteffizienzhypothese 5 2.2.2 Eregnisstudie 7 2.3 Annahmen 9 2.3.1 Datengrundlage 9 2.3.2 Thesenbildung 11 3 Grundlagen zur statistischen Auswertung 12 3.1 Renditeberechnung und –bereinigung 12 3.1.1 Stetige und diskrete Rendite 12 3.1.2 Marktbereinigte Renditen 13 3.1.3 Zeitpunktspezifischer t–Test 16 3.2 Definitionen 16 3.2.1 Stationarität, Leptokurtosis und Normalverteilung 17 3.2.2 Autokorrelation und Autokovarianz 18 3.2.3 White Noise, Random Walk und Lag–Operator 19 3.3 Linear stochastische Prozesse 20 3.3.1 Autoregressive Prozesse 21 3.3.2 Moving–Average–Prozesse 22 3.3.3 Autoregressive–Moving–Average–Prozesse 24 3.4 Modelle zum Volatilitätsclustering 25 3.4.1 ARCH–Modelle 25 3.4.2 GARCH–Modelle 27 3.4.3 Weiterführende Modelle 28 3.5 Berechnung erwarteter Gewinn 29 3.5.1 Hauptkomponentenanalyse 30 3.5.2 Saisonale Regression 31 3.5.3 Gewinnerwartung 33 3.6 Verfahren zu These 2 und 3 33 3.6.1 Koinzidenzanalyse 33 3.6.2 Bootstrapping 35 3.6.3 Monte–Carlo–Simulation 36 4 Literaturübersicht 37 4.1 Bisherige Studien über Gewinneinfluss 37 4.2 Studien mit Bezug auf ARCH–Modelle 38 4.3 Zusammenfassung 39 5 Ergebnisauswertung 40 5.1 Nachweis statistischer Eigenschaften 40 5.1.1 Stationarität, Leptokurtosis und Autokorrelation 40 5.1.2 Interpretation von Volatilität mittels GARCH–Prozess 44 5.1.3 Kritische Würdigung 47 5.2 Einfluss des Gewinns 47 5.2.1 Vorgehen 47 5.2.2 Ergebnisse 50 5.2.3 Kritische Würdigung 53 5.3 Ergebnisse der Koinzidenzanalyse 53 5.3.1 Vorgehen 53 5.3.2 Ergebnisse 56 5.3.3 Kritische Würdigung 58 6 Fazit 59