Do local proxy advisors matter? – Evidence from Germany

Abstract

Prior research documents that the large US-based proxy advisors, Institutional Shareholder Services (ISS) and Glass Lewis (GL), play an important role as information intermediaries in corporate governance worldwide. We provide initial evidence on the role of local proxy advisors, using the German setting. We analyse voting recommendations by local (IVOX) and foreign (ISS, GL) proxy advisors. First, we find that IVOX's voting recommendations differ substantially from those of ISS and GL. Second, we observe that IVOX's against-recommendations are significantly negatively associated with voting support. Third, we find that this association is particularly negative for voting outcomes at companies where local institutional investors hold larger stakes. Taken together, our findings suggest that the local proxy advisor IVOX considers relevant factors appreciated by local institutional investors that are distinct from factors incorporated in foreign proxy advisors’ voting recommendations.

Keywords:

• annual general meeting
• corporate governance
• institutional investors
• proxy advisor

Acknowledgements

We appreciate the insightful comments and suggestions we received from two anonymous reviewers, and the editor (Juan Manuel Garcia Lara). We thank ISS, Glass Lewis, and IVOX for providing the data. Furthermore, we thank two former members of IVOX for participating in interviews with the authors. We thank Beatrice Weder di Mauro for inspiring us to conduct this research. We have received valuable comments from Hans-Joachim Böcking, Oliver Entrop, Alexander Juschus, Devrimi Kaya, Thomas Loy, Nico Lehmann, Peter Limbach, Alexander Mosthaf, Jan Riepe, Bruno Rothacker, Anke Sänger-Zschorn, and Thorsten Sellhorn. We thank participants of the University of Mainz Brown Bag Seminar, the IAAER Accounting Conference 2018, the VHB Annual Conference 2018, and the MBF Rome Conference 2017. This study is partially based on chapters from the dissertation papers of the second and third author. We received funding for our project from the Research Priority Program Interdisciplinary Public Policy (IPP) at the JGU.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 ISS is the market leader, covering about 44,000 meetings each year and advising about 2000 institutional clients (ISS 2021). GL is the second largest, covering about 30,000 meetings each year and advising about 1300 institutional clients (GL 2021).

2 GL acquired IVOX in June 2015. As their integration took place at the end of 2015 (GL 2015), IVOX still used its own methodology for our full sample period (2013–2015).

3 Relatedly, Daines et al. (2010) find that governance ratings produced by proxy advisors do not provide useful information for shareholders.

4 Appendix II provides the interview guideline used for the two interviews, which were conducted in German. The first interview was conducted on 27 February 2020. The second interview was conducted on 13 March 2020. Both interviews were audiotaped and lasted around 60 minutes.

5 Katherine Rabin, the CEO of GL at that time, said: ‘The IVOX team's deep knowledge and experience analysing European companies will benefit Glass Lewis’ clients immediately, complementing the research being produced from our European headquarters in Limerick, Ireland’ (GL 2015).

6 Each of the three proxy advisors compiles reports providing some background information for each agenda item. We obtained access to the reports and recommendations of GL and IVOX and the recommendations of ISS.

7 We identify these companies using the CDAX,a stock index of the Deutsche Börse Group that includes all companies publicly listed at the regulated market at the Frankfurt stock exchange.

8 We access the digital version of the Hoppenstedt-Aktienführer from https://digi.bib.uni-mannheim.de/aktienf%C3%BChrer/.

9 Hitz and Lehmann (2018) provide information on ISS and GL coverage of a number of European countries for the sample period 2008–2010. During that period, ISS covered 46.8% and GL 22.4% of German firms, a lower proportion compared to our sample period for 2013–2015.

10 In other analyses (not presented), we find that performance measures (e.g. return on equity, stock market performance, sales growth) have no significant impact on the coverage decision for each of the proxy advisors. Further, when adding these measures as independent variables, the results for the other variables reported above remain robust. We do not include the performance measures in our main model for reasons of parsimony.

11 In additional tests, we analyse the determinants of coverage by only either the local proxy advisor IVOX (‘only local’) or the foreign proxy advisors ISS and GL (‘only foreign’). We find that only 29 firm-years are covered by the local proxy advisor alone, resulting in insignificant effects for the coefficients in the ‘only local’ determinants analysis using the specification in equation (1). We find that 188 firm-years are only covered by the foreign proxy advisors ISS and GL alone. A determinant analysis of ‘only foreign’ reveals that firms are less likely to be covered only by the foreign proxy advisors when the proportion of local institutional investors is larger. The coefficients of the other independent variables using the specification in equation (1) are insignificant.

12 In additional tests, we analyse the influence of foreign proxy advice in the absence of local proxy advisors to test for competition effects (e.g. Li 2018). However, the restricted sample includes only 80 firm-years of companies that are relatively small compared to the full sample (average market capitalisation of 0.23 bn Euro versus 6.75 bn Euro). Conducting the analysis shows that the proportion of against-recommendations of ISS and GL is similar but that the association of against-recommendations of ISS and GL with voting outcomes is weaker when comparing the results for the restricted sample to the full sample. One interpretation of this finding could be that ISS and GL issue less meaningful recommendations in the absence of IVOX. However, the small sample size involving relatively small firms requires a more cautious interpretation of this finding. Moreover, interviews with former IVOX leadership members reveal that proxy advisors have limited awareness about the firm coverage of other proxy advisors (R1, R2).

13 Hitz and Lehmann (2018) observe a slightly higher proportion of against-recommendation for ISS (10.5%) and GL (9.5%), using a different sample period (2008–2010) than we do (2013–2015).

14 One reason for the higher proportion of against-recommendations might be that IVOX, as a specialised local proxy advisor, evaluates agenda items more critically than foreign proxy advisors to differentiate itself. An alternative explanation might be that IVOX's local clients have specific preferences that drive the higher proportion of against-recommendations. In our interviews with former IVOX members, we learnt that the local client base of IVOX appreciated a more critical stance by the proxy advisor to protect their reputation in their home country. Local clients would often use against-recommendations as the starting point for conducting their own research. Further, local clients are interested to improve certain aspects of corporate governance in Germany, e.g. enhancing the transparency of corporate governance.

15 For reasons of conciseness, we include abstain-recommendations into the category of for-recommendations.

16 We define the voting outcome in line with prior literature. That is, the proportion of for-votes is scaled by the sum of for-votes and against-votes.

17 The results remain virtually unchanged when estimating Tobit models (Tobin 1958). As fixed effects estimations with Tobit models are more difficult to interpret, we report OLS results.

18 The results remain virtually unchanged for the restricted sample used in later analyses for testing differential effects of against-recommendations on voting outcome by origin of investors, which excludes observations with missing information on the shareholder structure (untabulated). Furthermore, our results remain robust when estimating separate regressions for each proxy advisor and restricting the sample to observations for which only the respective proxy advisor issues an against-recommendation.

19 In the previous analyses, we identify auditor election, director discharge, director election and executive discharge as agenda items where IVOX diverges substantially from ISS and GL. When we estimate separate regressions for items where IVOX diverges versus does not diverge much from ISS and GL, we find significant effects of IVOX against-recommendations on voting outcomes only for the subsample of items where IVOX diverges much. The effects of ISS and GL remain largely unchanged across both subsamples (untabulated). This finding is consistent with the idea that IVOX as the local proxy advisor matters most for agenda items that have local-specific features.

20 As the proportion of non-blockholders constitutes free float, we adjust the proportion of free float by subtracting the proportion of local and foreign non-blockholder institutional investors in models that include all those measures for ownership structure. Please note that we do not need to include the measures for ownership structure as main effects into the model because firm-year fixed effects already capture those factors. The results are robust for clustering standard errors at the firm level.

21 In the interviews with former IVOX members, we discuss whether IVOX integrated a stakeholder-orientation into its methodology. According to their responses, a stakeholder orientation would not matter at all (R1) or would only matter for agenda items related to M&A activities (R2).