How do ownership concentration and family control influence over-investment and financial risk?

The study employs a direct measure of over-investment introduced by Richardson (2006) to examine the association among ownership concentration, family control, and over-investment. We also highlight the moderating effect of family control on the relationship between free cash flow (FCF) and over-investment as well as the risks of over-investment. The estimation results for Taiwanese public-listed firms for the period of 2001–2014 indicate that over-investment is sensitive to FCF; however, the extent of over-investment of FCF is lower in the presence of family control. Our results clarify the influence of family control on over-investment and show that family controlled firms have lower (higher) levels of over-investment than their counterparts do when free cash flow is at high (low) levels.

Moreover, this study finds a non-linear relationship between ownership concentration and over-investment. An important implication of this research is that controlling shareholders may play either an expropriation or monitoring role. In line with the agency theory, concentrated ownership enables controlling shareholders to pursue their self-interests through over-investment. However, when ownership concentration is beyond an optimum level, controlling shareholders have less incentive to carry out sub-optimal investment decisions. Furthermore, we find that abnormal investment is related to financial risk in a U-shaped direction, indicating that both under-investment and over-investment cause higher financial risk. Our findings corroborate the need to account for non-linearity when investigating the influence of investment expenditures on financial risk. Overall, the present study fills the gap of determining the factors and consequences of over-investment by highlighting the role of family control and the problem of higher risk after over-investment. Various robustness checks, including alternative regression estimation models and different proxies, have been conducted to address the endogeneity problem using 2SLS and Fama–MacBeth regressions and the non-linearity issue of ownership-decision association.