The Ownership Concentration − Innovation Nexus: Evidence From SMEs Around The World

ABSTRACT

This study investigates whether ownership concentration improves or impedes firm innovation using a sample of small and medium-sized enterprises (SMEs) spanning 95 countries worldwide. We find that higher ownership concentration is associated with a lower likelihood of introducing innovative activities. Further, results reveal that concentrated ownership has detrimental impacts on innovation for firms with a higher degree of asymmetric information and firms led by less experienced managers. We also show that the negative association between ownership concentration and innovation only exists for financially constrained firms (i.e., younger enterprises and SMEs with high financing obstacles) and those with the highly concentrated ownership structure. Lastly, evidence suggests that institutional development alleviates the negative impact of ownership concentration on innovation.

KEYWORDS:

• Innovation
• SMEs
• ownership concentration

JEL:

• O31
• G32

Notes

1. See also Iwasaki and Mizobata (2020) for the comprehensive meta-analysis on the relationship between ownership concentration and firm performance.See also Iwasaki and Mizobata (2020) for the comprehensive meta-analysis on the relationship between ownership concentration and firm performance.

2. The existing literature does not offer a sharp distinction between the relationship between ownership concentration and innovation in general firms and that in SMEs. Our study, similarly to other papers focusing on SMEs (e.g. Deng, Hofman, and Newman 2013), employs the widely known theoretical background (e.g. the agency theory) to set up expectations on the impacts of ownership concentration on SMEs innovation.The existing literature does not offer a sharp distinction between the relationship between ownership concentration and innovation in general firms and that in SMEs. Our study, similarly to other papers focusing on SMEs (e.g. Deng, Hofman, and Newman 2013), employs the widely known theoretical background (e.g. the agency theory) to set up expectations on the impacts of ownership concentration on SMEs innovation.

3. We thank an anonymous referee for pointing out this channel.We thank an anonymous referee for pointing out this channel.

4. We appreciate an anonymous referee for suggesting this analysis.We appreciate an anonymous referee for suggesting this analysis.

5. “Soft innovation” refers to organizational innovation (i.e. new or significantly improved organizational or management practices and structures (Qi and Ongena 2019). We cannot capture this type of innovation using WBES due to the absence of questions about organizational improvement. Nonetheless, the application of “hard innovation” is beneficial to test a mechanism which refers to financing constraints since “hard innovation” is more capital-intensive than “soft innovation” (Qi and Ongena 2019).“Soft innovation” refers to organizational innovation (i.e. new or significantly improved organizational or management practices and structures (Qi and Ongena 2019). We cannot capture this type of innovation using WBES due to the absence of questions about organizational improvement. Nonetheless, the application of “hard innovation” is beneficial to test a mechanism which refers to financing constraints since “hard innovation” is more capital-intensive than “soft innovation” (Qi and Ongena 2019).

6. Unfortunately, WBES does not produce further information regarding the ownership structure such as the percentage of ownership held by the largest three or five shareholders as in Choi, Park, and Hong (2012); or by management or outside block holders as in Francis and Smith (1995).Unfortunately, WBES does not produce further information regarding the ownership structure such as the percentage of ownership held by the largest three or five shareholders as in Choi, Park, and Hong (2012); or by management or outside block holders as in Francis and Smith (1995).

7. We do not apply firm fixed effects due to some reasons as follows: (1) All firms engaging in the survey are in anonymity (i.e. without firm ID); (2) Ownership concentration changes slowly over time, so the use of firm fixed effect will compensate the effect of ownership concentration on innovation.We do not apply firm fixed effects due to some reasons as follows: (1) All firms engaging in the survey are in anonymity (i.e. without firm ID); (2) Ownership concentration changes slowly over time, so the use of firm fixed effect will compensate the effect of ownership concentration on innovation.

8. The WBES does not provide information on firms’ identity. Thus, we cannot have a standard panel dataset to proceed with lead-lag technique to alleviate reverse causality issue.The WBES does not provide information on firms’ identity. Thus, we cannot have a standard panel dataset to proceed with lead-lag technique to alleviate reverse causality issue.

9. The dataset is freely distributed at https://scholar.harvard.edu/shleifer/publications/quality-government. Information on legal origins is not available for seven countries: Democratic Republic of the Congo (DRC), Montenegro (MNE), West Bank and Gaza (PSE), Serbia (SRB), South Sudan (SSD), Timor-Leste (TLS), and Kosovo (XKX). In our sample, 26 countries have British legal origin; 34 countries have French legal origin; 27 countries have Socialist legal origin; 1 country has Scandinavian legal origin (Sweden); and no country has German legal origin. Thus, we have three dummies variables as instruments.The dataset is freely distributed at https://scholar.harvard.edu/shleifer/publications/quality-government. Information on legal origins is not available for seven countries: Democratic Republic of the Congo (DRC), Montenegro (MNE), West Bank and Gaza (PSE), Serbia (SRB), South Sudan (SSD), Timor-Leste (TLS), and Kosovo (XKX). In our sample, 26 countries have British legal origin; 34 countries have French legal origin; 27 countries have Socialist legal origin; 1 country has Scandinavian legal origin (Sweden); and no country has German legal origin. Thus, we have three dummies variables as instruments.

10. We obtain data on country-specific divorce rates, which covers 161 countries from 1920 to 2018, at https://ourworldindata.org/marriages-and-divorces. This website comprises data from various sources, including United Nation, OECD, and Eurostat. Since the statistics on marriage and divorce slowly change overtime, we replace missing values with the information available in the most recent year.We obtain data on country-specific divorce rates, which covers 161 countries from 1920 to 2018, at https://ourworldindata.org/marriages-and-divorces. This website comprises data from various sources, including United Nation, OECD, and Eurostat. Since the statistics on marriage and divorce slowly change overtime, we replace missing values with the information available in the most recent year.

11. We compute the median value of LN_EXPER for each country rather than the whole sample to account for differences in the educational system and culture among countries, which can determine the average age to engage in the workforce.We compute the median value of LN_EXPER for each country rather than the whole sample to account for differences in the educational system and culture among countries, which can determine the average age to engage in the workforce.

12. AUDIT is no longer a control variable when we conduct an analysis for firms with and without externally checked financial statements.AUDIT is no longer a control variable when we conduct an analysis for firms with and without externally checked financial statements.

Additional information

Funding

This work was supported by the Dalat University [DLU research fund.].