لینک کردن اعتبار شرکت ها و ارزش سهام با استفاده از انتشار رتبه بندی اعتبار

Good corporate reputation is one of the most valuable assets and causes a multitude of favorable impacts within different stakeholder groups. As a consequence, a lot of studies analyze the relationship between corporate reputation and financial performance. However, most of them raise the question of causation due to their methodology. In order to isolate the causal impact of corporate reputation on financial performance, some authors conduct event studies, but without remarkable success. Therefore, this study provides initially a comprehensive theoretical background for why reputation should affect financial performance. According to the resulting hypotheses, an event study is conducted to analyze the impact of the publication of reputation rankings by the German Manager Magazin on share prices. As hypothesized, positive and negative announcement effects exist regarding upgraded or downgraded companies respectively. Consequently, investors gain new information from the published rankings (increases or decreases in reputation) to adjust share prices.

The major change in management research during the last decades has been the paradigmatic move from thinking in tangible to intangible assets (Barney, 1991). This movement is attributable to the assumption that intangibles are the major drivers of sustainable performance due to their characteristics. Competitors cannot easily neutralize these assets because they are hard to copy and in general not tradable via factor markets. As one of these intangibles, corporate reputation has become one of the most discussed (Abimbola and Vallaster, 2007, Caruana, 1997 and Hunt and Morgan, 1995) and most valuable assets (Hall, 1992), because reputation is considerably able to defend a competitive position (Dierickx and Cool, 1989 and Jones et al., 2000) especially by buffering negative critical incidents (Dhir & Vinen, 2005). As a consequence, the assumption arises that a consistent and strong relationship between company reputation and financial performance exists. That implies that a relationship should also exist between information contained in corporate reputation rankings and financial performance. A number of research studies analyze this relationship (e.g., Dunbar and Schwalbach, 2000, Inglis et al., 2006, Rose and Thomsen, 2004 and Sánchez and Satorrío, 2007), but none of them confirms, beyond any doubt, an influence of corporate reputation on financial performance. Analyses either cannot prove the claimed effects or the direction of causation is not clear (McGuire et al., 1990 and McGuire et al., 1988). In order to isolate the effect of reputation on financial performance, Hannon and Milkovich (1996), Ittner and Larcker (1998), Fornell, Mithas, Morgeson, and Krishnan (2006) as well as Abraham, Friedman, Khan, and Skolnik (2008) look for announcement effects of publishing reputation data. However, these studies are unable to confirm an impact on shareholder value. Therefore, this event study is conducted using a refined methodology and different data (reputation rankings). More concretely, this study investigates whether announcing significant positive (negative) changes of corporate reputation measures affect shareholder value in the same direction. In contrast to previous ones, this study confirms a relationship as expected. The paper starts with theoretical bases of corporate reputation and shareholder value as well as their interdependencies. This passage includes an overview of general drivers derived from financial and management theory, followed by a closer look at the specific (possible) impact of good reputation on shareholder value. After that, a discussion of the information contained in published reputation rankings under the assumption of market efficiency follows. The next two sections describe the methodological approach and the used sample. Finally, the paper moves on to the presentation and discussion of the results.

The present analysis demonstrates two distinct results. On the one hand, publications of reputation rankings have an impact on shareholder value. As expected, a positive (negative) announcement effect exists if the relative ranking position has significantly improved (weakened) in comparison to the competitors. On the other hand, neither the information of good (bad) reputation scores in a ranking, nor their changes are solely appropriate to generate excess returns in the long run. Both results are in line with earlier argumentation. The announcement effect clearly indicates that corporate reputation is information which is not public. As a consequence, corporate reputation has to be disclosed. However, once published, share prices quickly fully reflect the information. Therefore, announcement effects are significant at the day of publication and limited to a small event window. This limitation and the dominance of other information in the long run support assumed market efficiency and the corresponding specification of short event windows. However, these findings contradict the empirical results of Hannon and Milkovich (1996), Ittner and Larcker (1998), Anderson and Smith (2006), Fornell et al. (2006) and Abraham et al. (2008). In order to clarify the underlying causes for these contradictions, a closer look at the similarities and differences between data and methodologies follows. First of all, to discover announcement effects, Hannon and Milkovich (1996), Ittner and Larcker (1998), Fornell et al. (2006) and Abraham et al. (2008) use rankings which differ in their conceptions to measure reputation, their number of publications and their publication cycle. But in principle, despite various research questions and respondents, all used rankings can potentially cause announcement effects due to the disclosure of non-public information. Only regarding the quarterly announcements of the American Customer Satisfaction Index (ACSI), data which Fornell et al. (2006) use, one can suppose that effects tend to be very small, attributable to the short time span between the announcements. Thus, widely enough differing expectations of investors are less likely to change buying intentions. The further discussion excludes the study of Ittner and Larcker (1998) due to the fact that Fornell et al. (2006) analyze more recent announcements of the same ranking and overcome some methodological weaknesses. In general, this study uses the same approach to test for announcement effects as Hannon and Milkovich (1996), Fornell et al. (2006) and Abraham et al. (2008) use. Nevertheless, considerable methodical differences exist between all studies which can lead to diverging results. The most important difference seems to be how the respective studies define events. These definitions are crucial to selecting and grouping companies. Hannon and Milkovich (1996) define the event as the publication of a human resource ranking. Consequently, they group and test all publically tradable companies in the ranking. Contrary to this, Abraham et al. (2008) define the event as listed in a specific quartile when the Reputation Quotient (RQ) is published. This definition takes into account the relative position of a company in comparison with competitors. Therefore, Abraham et al. (2008) group and test the companies in correspondence with their quartiles over all rankings. The problem of both event definitions is that they neglect the necessity of a changed (relative or absolute) position of companies. Without any change, a revaluation of share prices is not necessary for investors. In line with this argument, Fornell et al. (2006) define the event as the publication of changed ACSI scores. This means, comparable to this study, they analyze the impact of changed reputation scores in a ranking. However, Fornell et al. (2006) consider just an absolute change of scores, which is not a sufficient indicator for an improved or declined competitive position. Consequently, taking a dynamic perspective, this study considers relative changes in the reputation measure in comparison to both the previous ranking and the overall mean. A comparison of event window definitions shows that all studies use primarily the period of publication (τ = 0) and some additional event windows of maximum 5 units of time (τ = − 2 to τ = 2). Only Abraham et al. (2008) go beyond that and use an additional window of 250 days to compare their results with the study of Anderson and Smith (2006). The criticism towards such a long event window and, accordingly, towards Anderson and Smith's study (2006) is the same (see McWilliams & Siegel, 1997), which leads to a maximum of 3 days limit of event windows in this analysis. In all analyses the definitions of estimation windows differ in length (Hannon & Milkovich, 1996–12 or 6 months; Fornell et al., 2006–255 trading days; and Abraham et al., 2008 ─ 100 trading days) and regarding their positions in time. For example, Fornell et al. (2006) use a gap between the estimation and event window of 46 days. Whereas Abraham et al. (2008) use a gap of 100 trading days to ensure that the event itself does not influence estimated parameters. Even if the usage of such a gap is common practice, its completely arbitrary definition offers an opportunity to influence the results. This potential effect is partially visible when comparing the excess returns with the provided control window at τ = 0 (0.38 compared to 0.49). Rejecting consciously the use of any gap avoids such bias. With regard to various lengths of estimation windows, the impact seems negligible as long as the period is long (τ ≥ 100). Furthermore, variation in calculating actual ex-post returns is another potential cause for diverging results. Hannon and Milkovich (1996) and Fornell et al. (2006) use discrete returns. In contrast, Abraham et al. (2008) use log transformed returns, the same as this study. However, the usage of monthly returns by Hannon and Milkovich (1996) is fundamentally questionable to test for announcement effects. Finally, the isolation of an effect triggered by an event is almost impossible in such aggregated data. Another important factor, besides the event definition, is the selection of a market model. All the studies, including this one, used the classical CAPM as market model. However, in opposition to Hannon and Milkovich (1996) and Fornell et al. (2006), who estimated the parameter α as intercept, which corresponds with Rf,τ used here (see Eq. (6)), Abraham et al. (2008) set α to be zero. The β parameter, as a risk measure in comparison to the market, is estimated in all studies except the study of Abraham et al. (2008). Abraham et al. (2008) set β to be 1, which implies that all companies are as risky as the entire market. That is not in line with finance theory. To put it in a nutshell, the various used rankings are not attributable to the absence of event studies which confirm announcement effects when reputation rankings are published, but rather methodological differences. Nevertheless, in order to show an effect resulting from corporate reputation, the data should fulfill some basic criteria. The data should not be colluded due to sector membership of respondents and not be biased because of financially focused criteria (Bromley, 2002a). Manager Magazin's data fulfill both criteria ( Schwaiger, 2004). This also holds true for ACSI data used by Fornell et al. (2006) and all reputation rankings used by Hannon and Milkovich (1996). The suitability of RQ data ( Abraham et al., 2008) is questionable ( Schwaiger, 2004), and the Fortune data used by Anderson and Smith (2006) are inappropriate with respect to these criteria ( Bromley, 2002a).