Does Managerial Foreign Experience Deter Corporate Fraud

ABSTRACT

We explore how managers with foreign experience affect corporate fraud in China. By employing a bivariate probit model with partial observability, we find that returnee managers significantly reduce the incidence of corporate fraud, and increase the probability of being detected, dependent on the given fraud. Improved corporate information environment may mainly drive our results. Furthermore, the impact of returnee managers on fraud deterrence also varies according to the different nature of foreign managerial experience, positions of returnee managers, and types of corporate frauds. Overall, we offer new evidence that returnee managers have an increased awareness of corporate fraud.

KEYWORDS:

• Managerial foreign experience
• corporate fraud
• information environment
• China

Acknowledgments

We gratefully acknowledge the financial support from the National Natural Science Foundation of China [71902185]; Fundamental Research Funds for the Central Universities, Zhongnan University of Economics and Law [2722021BZ054]; All errors are our own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. https://www.caixinglobal.com/2020-05-20/exclusive-luckin-prepares-for-the-worst-in-face-of-delisting-risk-101556097.html.

2. https://www.globaltimes.cn/content/1157538.shtml.

3. Given the scarcity of managerial talents in emerging markets like China, it is suggested that returnee managers are considered as enterprise super stars within the capital market, and once back in China they have an “eyeball effect”, receiving more monitoring and attention from government authorities, stakeholders, financial analysts, auditors, institutional investors, as well as the public (Yuan and Wen 2018).

4. The sample period started from 2008 because the CSMAR database only provided information about managers’ foreign experience since 2008.

5. The bivariate probit model theoretically provides an appropriate technique to address partial observability issues. However, it still encounters practical difficulties with poor convergence attribute (Firth, Rui, and Wu 2011; Khanna, Kim, and Lu 2015). As indicated in Khanna, Kim, and Lu (2015)’s paper, including two-way fixed effects may result in a no-convergence problem, and they add only year fixed effect in the baseline model. We also face similar challenges in our paper, we thus acknowledge this weakness for bivariate probit model and include some industry-level variables to partly handle this problem following Khanna, Kim, and Lu (2015).

6. For instance, if an enforcement action filed by CSRC in 2016 states that a firm i once committed frauds in 2008, 2010, and 2015, we define the firm-year observations of 2008, 2010, and 2015 as fraud sample and denote Z_i,2008, Z_i,2010, Z_i,2015 equal to 1. After checking all the enforcement actions, we then include the firm-year observations which never experience an incidence of fraud in the non-fraud sample and denote them Z_it = 0.

7. Appendix B also presents the sample distribution of Chinese fraudulent firms by year (Panel A), by industry (Panel B), and by fraud types (Panel C). The year of fraud denotes the year when a fraud event is committed, rather than the year of its detection. In Panel A, we find that over the sample period, the proportion of firms accused of fraud accounts on average for 15.18% in full sample, peaks in year 2012 (=19.93%), and over time remains relatively stable at around 15%. In Panel B, we describe the number of firm-year observations, the firm-year observations with fraudulent activities, as well as the proportion of frauds in each industry. It shows that frauds are more frequent in industries such as computing and communications equipment (346) and chemical materials manufacturing (325). However, industries such as waste resource utilization (33.33%), education (28.57%), wood processing (24.64%), and petroleum processing (24.31%), have a relatively high proportion of frauds. In Panel C, we present the list of fraud types as classified by CSRC and find the most common frauds are delayed disclosures (19.73%), major information omissions (16.65%), false records (misrepresentations) (13.15%) and other types (27.91%). The 3927 fraud observations describe 8822 fraud types, demonstrating that some firms usually have multiple types of violation in a given year.

8. We use P(F = 1) and P(D = 1|F = 1) for brevity in the following sections.

9. According to Yang and Ye (1993), the colonies or concessions of Britain include Guangzhou in Guangdong province, Hankou in Hubei province, Jiujiang in Jiangxi province, Weihai in Shandong province, Xiamen in Fujian province, Zhenjiang in Jiangsu province, and two municipalities, Tianjin and Shanghai.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [71902185]; Fundamental Research Funds for the Central Universities, Zhongnan University of Economics and Law [2722021BZ054].