Hong Kong stock listing and the sensitivity of managerial compensation to firm performance in state-controlled Chinese firms

State-controlled Chinese firm executives often enjoy significant perks commensurate with their job titles (Cai et al. 2005). We do not include perks in the annual cash compensation because data on perks are not readily available. The provision of perks could be a form of incentive pay if an executive’s turnover is sensitive to firm performance. As shown later in Table 5 , we find little evidence that state-controlled Chinese firms’ CEO turnover is sensitive to firm performance. Therefore, it seems unlikely that omitting perks would significantly affect our inferences.

See Karolyi (2006) for a comprehensive survey of this literature.

See Appendix VII of the China Construction Bank’s Prospectus for a detailed discussion of these provisions (available at http://​www.​hkexnews.​hk/​listedco/​listconews/​sehk/​20051014/​939/​F135_​e.​pdf).

Hong Kong returned to China in sovereignty on July 1, 1997, after 100 years of British rule. In accordance with the "One Country, Two Systems" principle agreed between the UK and China, Hong Kong's previous capitalist system and its way of life were promised to be unchanged for 50 years.

Anecdotal news reports suggest that the influence of the Chinese government is still prevalent in the management of state-controlled H shares and Red Chip shares, especially with regard to the appointment of top executives. For example, on November 5, 2004, the Chinese government suddenly swapped top executives at China’s three top telecommunications companies: China Mobile (Red Chip), China Unicom (Red Chip), and China Telecom (H share). Likewise, the retirement of the chairman of CITIC (unlisted), one of China’s largest state-owned conglomerates, triggered top level management reshuffles at two major mainland banks in July 2006, China Construction Bank (H share) and Bank of Communications (H share).

A unique feature of state-controlled Chinese firms is that a CEO who performs well in a state-controlled firm (say X) may be promoted to a higher-ranking government position (for example, becoming a minister) or transferred to another more prestigious state-controlled firm (say Y) by the government, similar to a tournament (see Rosen 1986). Although such a CEO turnover may be viewed as a reward to the departing CEO of firm X, it does not maximize the shareholder value of firm X: firm X loses a good CEO if that CEO performs well, but firm X ends up keeping a bad one if he performs poorly. Therefore, though CEO turnover is sensitive to firm performance, the sensitivity is positive and thus is not consistent with shareholder value maximization. For this reason, we should not exclude such CEO turnover from our analysis.

ROA may not be comparable across the three types of firms because A shares and Hong Kong listed firms follow different financial reporting standards (local Chinese GAAP versus IAS). However, Chen et al.’s (2002) analyses suggest that the Chinese accounting standards in our sample period are largely harmonized with the accounting standards under IAS. Therefore, it seems unlikely that differences in accounting standards across the three types of firms are a significant issue in our study. Another concern with ROA is that earnings management could be more aggressive and thus the quality of ROA could be lower in state-controlled A share firms than in state-controlled H or Red Chip firms. Therefore, one would naturally expect the optimal weight on ROA in the cash compensation contract to be lower while the weight on alternative performance measures (for example, RET) to be higher in state-controlled A share firms than in state-controlled H or Red Chip firms (see Engel et al. 2003). Alternatively, one may also expect shareholders to turn to alternative compensation contracts (for example, long-term financial incentives) in order to motivate managers to increase shareholder value. Inconsistent with these alternative explanations, we find no evidence that the coefficient on RET is larger for state-controlled A shares in Table 2 or the level of long-term managerial incentives is higher for state-controlled A shares in Table 4 .

Inference is similar if EQUITYOWN is defined using Core and Guay’s (1999) approach, defined as the dollar value change in managerial stock and option ownership to a 1% change in stock price.

As EQUITYOWN includes managerial share ownership, regression model (3) implicitly assumes that shareholder value monotonically increases with managerial share ownership. However, this may not be true if the managerial share ownership becomes so high that the managers become entrenched. The evidence in Morck et al. (1988) suggests that the managerial entrenchment effect may exist when the managerial share ownership exceeds 5%. As managerial equity ownership for our sample firms rarely exceeds 5%, we do not believe that the entrenchment effect is a concern for our study.

With the presence of a controlling shareholder, the managerial pay-for-performance sensitivity for cash and equity compensation could be reduced because the controlling shareholder has the power to terminate the underperforming CEO (see Ke et al. 1999). However, we do not believe that the pay-for-performance sensitivity for cash and equity compensation will drop to zero. Furthermore, the controlling shareholder in our case is the Chinese government, and therefore it is unlikely that its monitoring is as intensive as a private controlling shareholder (see Shleifer 1998). Inferences are qualitatively similar if the coefficients on ROA and RET in model (1) are allowed to vary with PARENT_OWN or if PARENT_OWN is added to model (3).

We assume that the top executives of Red Chip firms live in Hong Kong. This may not be true because these firms have the majority of their business in China, and thus the executives could reside in mainland China for part of the year. However, excluding LN(COSTOFLIVING) from regression model (1) does not alter our inference.

The 13% for A shares is computed as e^{(2.274*SDroa−0.056*SDret)}−1, where SDroa and SDret are A shares’ standard deviation of ROA and RET, respectively. The 47% for H shares is computed as e^{((2.274+1.938)*SDroa+(−0.056+0.036)*SDret)}−1, where SDroa and SDret are H shares’ standard deviation of ROA and RET, respectively. The 80% for Red Chip shares is computed in a similar fashion.

Consistent with the managerial power theory (see Finkelstein 1992; Lambert et al. 1993), we find that LN(CASHPAY) increases (decreases) with the CEO’s stock ownership (the controlling shareholder’s stock ownership). However, including these two control variables in model (1) does not alter our inference in Table 2.

Results are similar if we exclude the CEO’s stock ownership from LONGTERMINCENTIVE and EQUITYOWN, suggesting that our results are not entirely driven by managerial share ownership alone.