| dc.description.abstract | In this thesis, we study primary insider holdings' effect on firm performance. The objective is to shed light on the complex relationship between corporate governance mechanisms and firm performance, researching both the ownership structure and identity dimensions in the same theoretical framework. This is done to account for the internal conflict between shareholders.
We build on the model of Demsetz & Vilalonga (2001) using pooled OLS, fixed-effects and two-stage least squares regression analysis. Our choice of models provides us with robust estimates and mitigates the risk of bias due to omitted variables, enabling us to compare the results with different econometric approaches. We use a rich dataset of firms with a primary listing at the Oslo Stock Exchange from 2010-2017. Additionally, we introduce insider liquidity as an instrument for primary insider holdings, which to our knowledge has not been done before with data from the Oslo Stock Exchange.
Our findings suggest that the amount of primary insider shares held by individuals does not impact firm performance. Primary insider holdings are only significant when using two-stage least square estimation regression analysis, and the significance is dependent on the instrument used. These findings suggest that ownership characteristics are of little significance to firm performance. This is consistent with Demsetz (1983) equilibrium hypothesis. Still, the lack of significant results might be explained by weak instruments. We conclude that until there is a stronger theoretical framework in place, the simultaneous equations approach remains ambiguous.
Furthermore, we find that firm size and turnover are the most consistently significant factors for Tobin's Q, which corroborates our theoretical framework and previous research. We also find that our results, from regressions done with data from the OSE, are in line with results from the U.S. and the U.K. This suggests that the prevalent pooled OLS, fixed-effects and two-stage least square regression models are independent of the named country's regulatory frameworks. | nb_NO |
