Empirical Methodology. To test this hypothesis, we could perform a cross-sectional comparison of the pre-repurchase agreement parameters of repurchasing firms, and relate these parameters to the repurchase decisions cross-sectionally. However, there is an endogeneity problem in that the agreement parameters and the decision to repurchase may have both been driven by some unobserved intrafirm variation. Specifically, firms vary significantly in their payout decisions, payout modes and magnitudes of cash distributions, and these differences persist over time. The literature has shown that large and mature firms are more likely to distribute cash through repurchases or dividends or both relative to small and young firms. Moreover, institutional investors form clienteles in terms of the types of firms they wish to invest in. For instance, prudence considerations lead institutional investors to prefer firms with large market capitalizations, sufficient liquidity and low volatility (Del Xxxxxxx (1996) and Xxxxxxx and Xxxxxxx (2001)). Finally, analysts’ earnings forecast dispersions also vary across different types of firms, typically being lower for low-growth and mature firms than for high-growth firms. Thus, if the same unobservable time-invariant factors that determine investor clienteles and cross-sectional variations in analysts’ forecast dispersions are also those that influence firms’ repurchase decisions, then there is an identification problem. Ignoring this problem may lead us to erroneously conclude that a firm with higher institutional ownership and lower analysts’ EPS forecast dispersion is more likely to undertake a repurchase. To tackle this identification problem and mitigate the concern about potentially omitted variables, we rely on firm fixed-effect regressions and exploit within-firm variations while controlling for any time effects, similar to Khwaja and Mian (2008).23 With this firm fixed-effect approach, the effect of any unobserved firm-specific factors that may have driven both the agreement level and the decision to repurchase is completely absorbed since we make the comparison across different time periods for the same firm. Specifically, the variables of interest like the repurchase decision and the level of manager-investor agreement, as well as other control variables, are time-demeaned for a given firm, and they should be interpreted as the deviation from the average level for the firm. For instance, we interpret a drop in institutional ownership of a fi...
Empirical Methodology. To estimate the impact of Chinese imports on the productivity, profitability and innovation activity of these firms, the following models were adopted:
