- Original authors / creators
- Jia Liao; Yun Zhan; Yu Yuan; Ailing Xu
- Original publication
- PLOS ONE · Public Library of Science · 2024-07-31
- License & retained notices
- CC BY 4.0
© 2024 Liao et al
Article-specific license evidence ↗Exact license notice
This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Article-specific license URL: https://creativecommons.org/licenses/by/4.0/ - Edition changes
- Attributed narrative text edition with normalized whitespace. Original full XML retained. Figures, table layouts, display equations and supplements are available through the linked original full text; no new primary findings or source-author endorsement are claimed.
- Evidence limits
- Adjacent research, not a directly sampled family-office study. Applicability to a family office requires separate assessment. We conduct the 2SLS regression to tackle endogeneity due to reverse causality and omitted variables. Referring to the idea of constructing instrumental variables in the existing literature [60,61], we select the following two instrumental variables: (a) the percentage of firms with NCLSs in the industry to which the firm belongs (IV1), (b) the percentage of firms with NCLSs in the province where the firm is located (IV2). Table 11 shows the results of 2SLS regression, from the first stage estimation results in column (1), column (3), and column (5), both the regression coefficients of Dev×IV1 and Dev×IV2 are significantly positive at the 1% level. The results of the validity tests of the instrumental variables are as follows: (a) The F-statistics of the first stage are 246.63, 202.06, and 214.03. (b) The Anderson canon. corr. LM statistics are 483.358, 397.554, and 420.666, rejecting the null hypothesis of the underidentification test. (c) The Cragg-Donald Wald F statistics are 246.631, 202.061, 214.032, rejecting the null hypothesis of the weak identification test. (d) The P-values of the Sargan statistics are 0.1482, 0.8759, and 0.1914, indicating that both IV1 and IV2 are exogenous. From the second-stage estimation results in column (2), column (4), and column (6) of Table 11, we find that the regression coefficients of Dev are always significantly positive at the 1% level, and the regression coefficients of Dev×Dum, Dev×Num, and Dev×Ratio are 0.387, 0.232, and 2.068, respectively, and all are significant at the 1% level, consistent with previous results.
- Retraction / correction check
- No notice found in retrieved article XML, publisher HTML or Crossref relation/update-to/updated-by. Bounded check; unlinked or unregistered notices and independent retraction databases are outside coverage. Verified 2026-11-05. Automated/source checks are not human peer review.
- Preserved record
- Rights and full text checked 2026-11-05. Included in this volume on 2026-11-05. Original source dates remain unchanged.
Download the preserved source
SHA-256 (retrieved artifact): 6c222ab96ce608567663ede19d59a9b2dcb15540bc07776cfcd68f6600592106
SHA-256 (preserved text edition): cb040b1f20226022550e54631c1c03e62719ca4791ff8bfeea71f8afbb89daf5