Chinese state-owned enterprises (SOEs) have become quite profitable recently. As the largest shareholder, the state has not asked SOEs to pay dividends in the past. Therefore, some have suggested that the state should ask SOEs to pay dividends. Indeed, the Chinese government has adopted this policy advice and started to demand back dividend payments starting from 2008. Although we do not question the soundness of the dividend policy, the point we raise is whether those profits are real if all costs owed by SOEs are properly accounted for. Among others, we are interested in investigating whether the profits of SOEs are still as large as they claim if they were to pay a market interest rate. Using a representative sample of corporate China, we find that the costs of financing for SOEs are significantly lower than for other companies after controlling for some fundamental factors for profitability and individual firm characteristics. In addition, our estimates show that if SOEs were to pay a market interest rate, their existing profits would be entirely wiped out. Our findings suggest that SOEs are still benefiting from credit subsidies, and they are not yet subject to the market interest rates. In an environment where credit rights are not fully respected, dividend policy, though important, should come second and not first.

This paper examines the information content of firm ratings. We disentangle the relative contribution to firms' ratings of sovereign risks and of the individual firms' performance indicators employed by rating agencies. We reach three conclusions. First, the contribution of sovereign risk to firm ratings is high in developing countries but is negligible in developed countries. Second, even after controlling for the “country ceiling effect” (i.e., the constraint put on the private firms' rating by the rating of the country in which the firms operate), the information content of ratings for firms in developing countries is much smaller than for firms in developed countries. Third, cross-country indicators of information quality help explain these discrepancies, but they do not entirely account for them.