Essays on Accounting Information Quality in China

This research contributes to provide a better understanding of the nature of accounting information reliability by measuring the relation between the informativeness of earnings and corporate governance based on the Chinese context with its unique political, social, cultural and economic environment and large sample size. In particular, mainland China has a distinct two-tier board structure comprising a supervisor board including employee representatives and board of directors of whom at least one third are independent directors. The objective of this thesis is to investigate accounting information reliability and corporate governance by addressing three predominant empirical research questions in three studies. The first study examines the impact of board composition and independence on earnings management in mainland China through investigating whether independent directors and supervisors are effective at restraining earnings management. To fully capture the earnings attributes, the second study investigates the quality of reported earnings in China from the perspective of both accounting-based (including accrual quality, persistence, predictability and smoothness) and market-based earnings attributes (including value relevance, timeliness, and conservatism and earnings response coefficient). A two-way test has been conducted to compare the difference in earnings quality between State-Owned and Non-State-Owned enterprises. According to financial distress theory, the incentives for Non-SOEs to manipulate earnings are stronger than in SOEs, since SOEs have the advantage to receive financial subsidies from government while Non-SOEs face more financing constraints. The agency theory, however, argues that state ownership in SOEs creates incentives and regulatory backing for self-serving purposes, thus motivating SOEs to manipulate accounting numbers. The political cost hypothesis complements the agency theory and illustrates that SOEs’ managers would manipulate accounting numbers in response to government intervention (report conservatively to disguise the profits or report aggressively to meet specific thresholds). In addition, it tests whether analysts' forecasts are more accurate than forecasts based on time-series predicted statistics with random walk. Finally, the third empirical study detects whether managers intend to manipulate earnings via discretionary accruals in order to just meet or beat consensus analyst forecasts on the basis of earnings surprise (analyst forecast error). The key findings of the first study in this thesis suggest that the distinct Chinese two-tier board structure comprising independent directors and supervisory directors fails to mitigate earnings management. The second study documents that Chinese SOEs overall exhibit a lower earnings quality than Non-SOEs, supporting the agency theory. Government ownership might create incentives and regulatory backing for self-serving purposes that negatively influence the listed firms’ financial reporting. Moreover, SOEs manipulate downwards the earnings much more than Non-SOEs, manifesting the government generally expropriate the benefits of SOEs, according to the political cost hypothesis. One interesting finding in second study is that predicted earnings based on the time-series statistical model with drift are more accurate than the consensus analyst forecast. This result conflicts with findings from developed country studies, indicating the malfunction of financial analysts in mainland China. In the third empirical study, the findings suggest an optimistic bias in analysts' forecasts exists in Chinese listed companies but fail to provide any evidence supporting that discretionary accrual measures are positively associated with just meeting or beating the analysts’ forecast benchmark. It challenges the ‘benchmark beating’ incentive in most prior literature based on western developed countries, such as the US and the UK