Corporate Philanthropy and Tunneling: Evidence from China

It is also known as type II agency problem.

Our study shows that philanthropy is negatively associated with tunneling. However, we are not suggesting that the cost of donation is lower than the benefits. Firms always need to make a trade-off when deciding whether to invest in philanthropic activities (McWilliams and Siegel 2001).

We thank the reviewer for suggesting RELIGION as the instrument variable. Apart from 2SLS, we perform lead-lag change analysis and change model to assuage the endogeneity concerns. The results are unchanged.

We find similar results when we measure RELIGION as the number of nation-widely famous Buddhist monasteries and Taoist temples within a 200 km radius around the firm’s registered addresses.

Our sample starts in 2003, when firms began to disclose donation information in annual reports. Following Tu and Yu (2014) and Jiang et al. (2010), our sample ends in 2006, before which year other receivables were used extensively by controlling shareholders to transfer assets and profits out of listed companies.

Our results are complementary to the findings of previous literature (e.g., Li and Zhang 2010). What we find in our paper is that the negative association between philanthropy and tunneling is stronger in non-SOEs, suggesting that philanthropic donations by non-SOEs are more strategic because these firms have more incentives to enhance public image. Our findings suggest that investors should take ownership type into consideration when they assess the consequence of corporate philanthropy. In the same vein, Li and Zhang (2010) find that corporate ownership dispersion is positively associated with corporate social responsibility (CSR) for non-SOEs while negatively associated with CSR for SOEs. Their study alerts investors to distinguish SOEs from non-SOEs when evaluating firms’ CSR behavior.

We thank the reviewer for the suggestion of conducting lead-lag change analysis.

We thank the reviewer for pointing this out.

Asset acquisitions are transactions that involve the acquisition of tangible or intangible assets by the listed company from a connected person or from a private company majority-controlled by this person and calculated as the total acquisition amount divided by the total assets. Asset sales are transactions that involve the sale of tangible or intangible assets by the listed company to a connected person or to a private company majority-controlled by this person and calculated as the total sales amount divided by the total assets. Equity sales are transactions that involve the sale of an equity stake in the listed company to a connected person or a private company majority-controlled by this person and calculated as the total sales of equity stakes divided by the total assets. Trading relationships are transactions that involve the trade of goods and services between the listed company and a private company majority-controlled by a connected person and calculated as the total trading amount divided by the total assets. They can be purchases by the listed company or sales or both. (Cheung et al. 2006).

To mitigate the concern that the four ratios have different ranges and variations and are not appropriate to aggregate them directly, we further used the following methods to improve the robustness: (1) regress on the four ratios separately; (2) calculate the fractional rank for the four ratios first and then sum up. When we regress on the four ratios separately, we found that corporate philanthropy is only significantly negative when related party transaction is trading relationships. This alleviates the aforementioned concern that firms doing corporate philanthropy turn to more implicit type of tunneling. When we use the sum of fractional rank of the four ratios, we get qualitatively same results. To keep brevity, these results are untabulated.

We also use CP_SUM3 (calculated as the number of times a firm makes donations in year t − 2, t − 1, and t) to measure SUS, and our main results still hold.

We use the period of [−2, +2] because the China Securities Regulatory Commission requires listed companies to disclose information in no more than two trading days from the transaction day. To increase robustness, we also use the period of [−1, +1] and [−3, +3]. The empirical results remain similar.