Family Ownership and Impression Management: An Integrated Approach

ABSTRACT

According to agency theory, impression management is lower in family-owned firms because of reduced agency I conflicts. However, when family owners are present, agency I conflicts are superseded by agency II conflicts between controlling and minority shareholders, leading to a positive association between impression management and family ownership. Moreover, family owners have distinct goals related to the preservation of socioemotional wealth (SEW) that may influence agency conflicts. We formalize these countervailing forces by developing an integrated theoretical framework that considers both agency I and agency II conflicts and how they interact with SEW-related motives. We hypothesize that the coexistence of agency I and agency II conflicts in family-owned firms leads to a U-shaped association between family ownership and impression management. We also expect that this U-shaped association is flatter when family influence on the board is stronger and SEW motives shape more corporate choices. We test our predictions by analyzing the letters to shareholders of 77 Italian-listed firms from 2008 to 2015. The results confirm our expectations and hold when we use instrumental variable estimation, alternative proxies to capture SEW motives, and impression management measures.

Keywords:

• Family Ownership
• Impression Management
• Agency Theory
• Socioemotional Wealth
• Family Involvement

Acknowledgments

We appreciate helpful comments from the editor (Beatriz Garcia Osma), two anonymous reviewers, Cristina Cruz, Carlo D’Augusta, Marta Geletkanycz, Enrico Laghi, Daniele Macciocchi, Luis Gomez-Mejja, Annalisa Prencipe, Carlo Salvato, Marco Trombetta, Alessandro Zattoni, and the participants in the 2015 EAA Congress in Glasgow. We thank Emma Moretti and Arianna Pisciella for excellent research assistance. Saverio Bozzolan gratefully acknowledges the financial support of the minister dell’ Università e della Ricerca (PRIN 2009); and Marco Mattei gratefully acknowledges the financial contribution from Sapienza Università di Roma (Project 2013 no. C26A13NW2N). Part of this research has been conducted when Saverio Bozzolan was at the Department of Economics and Business (University of Padova) and Marco Mattei at the Department of Law and economics of production activities, Sapienza Università di Roma.

Disclosure statement

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

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, https://doi.org/10.1080/09638180.2022.2141287.

Appendix A: List of keywords

Appendix B: Manual coding procedure for the computation of impression management proxies

Appendix C: Variable definitions

Notes

1 We describe our proxies for impression management in the 3.3 section.

2 In our sample, the percentage of firms with an English version of an LTS is lower than the percentage of firms with an Italian version of an LTS. Moreover, when present, the English version is not original, rather it is a translation of the original Italian document. Previous studies (Campbell et al., 2005; Courtis & Hassan, 2002) document that that the translation of annual reports results in statements that are more difficult to read and interpret. Moreover, there is also evidence that the publication of an English annual report in non-English countries is associated with firm characteristics such as size and level internationalization, thus introducing potential selection bias in the sample (Jeanjean et al., 2010). Accordingly, we have decided to focus on the Italian versions of LTSs.

3 An alternative procedure is a computer-based content analysis to enlarge the sample size and reduce subjectivity concerns. Yet, as we focus on a single country like Italy, the number of firms to consider is limited ex ante.

4 Following Weber (1990), we define a statement as ‘a cluster of words with different meanings or connotations that, taken together, refer to some theme or issue’ (p. 37).

5 According to Italian law (D. Lgs. 58/1998, art. 106), although it is not a majority stake, 30% of the capital represents a controlling interest, which allows the owner to appoint members to the board of directors, significantly interfering in the management of the company and conferring the power to promote or block more or less all business decisions. For this reason, minority shareholders must be able to easily sell their shares if they do not favor a change of leadership or any type of related interference.

6 As we are interested in the co-existence and interplay of agency dynamics and SEW motives, it is crucial to assure that family owners can influence both, and differences arise as a result of the diverse SEW levels. Thus, at the moment of testing H2, we exclude firms with no family involvement in the board.

7 In additional analyses, we use Family_CEO and Family_boad_presence as separate conditioning variables. The results are in line with H2: the U-shaped association arises when the CEO is not part of the family or the family board influence is below the median. The relation is flatter when the CEO is a family member or the family board influence is above the median. Moreover, we use the average tenure of board members as an additional board feature shaping the effectiveness of board monitoring. According to an agency perspective, when the average board tenure is greater, board members are more likely to be entrenched and, hence, less able to curb managers’ opportunistic behaviours. In line with this idea, we find that the U-shaped relationship between impression management and family ownership is steeper when average board tenure is below the mean.

8 The evidence that the relationship between impression management and family ownership varies with market conditions in a predictable way also mitigates concerns that our results are due to unobservable firm characteristics.

9 Our instrument meets both conditions to be valid. First, it is relevant because, when the firstborn is a male, family owners have strong legacy incentives and, hence, a strong desire to keep the control to pass the firm to the firstborn male child. It also meets the exclusion restriction as it affects impression management only through family ownership (i.e. there is no direct linkage between firstborn gender and tone). As we instrument both Famown and its squared term, we also add the number of children alone and interacted with Firstborn in the first step.

10 To understand the procedure that leads to our dummy variable for the gender of a family owner’s firstborn child, consider a public corporation traded on the Milan Stock Exchange for which an LTS is available in our sample: Luxottica S.p.A. CONSOB reports that its largest shareholder is Leonardo Del Vecchio. In 2009, he held 69% of the company’s shares. Searching Google for ‘Leonardo Del Vecchio figli,’ we find that this individual has six children (Claudio, Marisa, Paola, Leonardo Maria, Luca, and Clemente), of whom four are male and two are female. The eldest one, Claudio, is a male and was born before 2008 (i.e., when our data collection starts). As a result, Firstborn takes a value of 1 in 2009 for Luxottica. As Leonardo Del Vecchio is the main shareholder of Luxottica for the whole sample period, Firstborn is equal to 1 for the whole sample period as well.

11 We ensure the validity of the coding process, a different coder applied the procedure to collect information on firstborn child to a random subset of 14 firms (representing 23.3% of the sample). The Pearson correlation among the raters is 96% (with only one case of potential disagreement). Moreover, the coder inspected all the cases for which the indicator was not available. The coder confirmed that the indicator could not be determined as there were not enough information on the number of children, their gender and, above all, age to determine the firstborn. This ensured a low probability of Type I error.

12 We check the validity of the instrument, running the Hansen test (Hansen & Singleton, 1982). In both cases of low and strong board influence, the null hypothesis cannot be rejected so we conclude that the instrument is valid.

Additional information

Funding

This work was supported by Sapienza Università di Roma: [grant no Project 2013 no. C26A13NW2N]; Ministero dell'Università e della Ricerca: [grant no PRIN 2009].