Matching method | PSM | Bootstrap (50) | Bootstrap (100) |
|---|
Nearest neighborhood | −0.0154*** | − 0.0154*** | − 0.0154*** |
|  | (−5.24) | (−5.23) | (−4.19) |
| Â | [0.97] | [0.00] | [0.00] |
Gaussian kernel | −0.0154*** | −0.0154*** | − 0.0154*** |
|  | (−6.08) | (−5.28) | (−5.71) |
| Â | [1.00] | [0.00] | [0.00] |
Radius caliper matching | −0.0147*** | −0.0147*** | − 0.0147*** |
|  | (−4.92) | (−3.19) | (−3.74) |
| Â | [0.81] | [0.00] | [0.00] |
Local linear regression | −0.0149*** | −0.0149*** | − 0.0149*** |
|  | (−4.46) | (−6.41) | (−5.34) |
| Â | [0.57] | [0.00] | [0.00] |
- This table presents the tests of differences in performance between non-family-chair firms and the corresponding matched family-chair firms. We match each non-family-chair firm with a family-chair firm using the one or two nearest neighborhood matching, Gaussian kernel matching, radius matching with a caliper of 0.0001 and matching with local linear regression. The variables in matching are the same as those in the main regression, including firm size, market-to-book ratio, leverage, tangibility, firm age, capital expenditure and a set of board and ownership characteristics. Year and industry dummies are also included. Column 1 reports the average treatment effects on the treated (ATT) for various matching processes when the treated group is non-family-chair firms. Columns 2 and 3 show the results when we use bootstrapped standard errors based on 50 and 100 replications. The t-statistics are in parentheses. Values in brackets are p-values for balancing tests in column 1 and for t-statistics in columns 2 and 3. Asterisks denote statistical significance at the 1% (***), 5% (**) or 10% (*) level, respectively
