Hai Zhang

We develop a dynamic trade-off model to explore how shareholders' risk-taking incentives interact with firms' debt structure choices over the business cycle. The model introduces a novel channel-risk-taking-that drives the heterogeneity of debt structures. Macroeconomic risk amplifies the sensitivity of debt structure decisions to these risk-taking incentives. The use of both bank and market debt exacerbates the ex-post risk-taking problem compared to relying solely on market debt, particularly under larger recessionary shocks. Firms optimally increase their reliance on bank debt when facing lower systematic risk exposure, when confronted with larger recessionary shocks, or when macroeconomic conditions deteriorate. Finally, we find that macroeconomic risk is a critical factor in shaping the relationship between shareholders' bargaining power and firms' optimal debt structure decisions.

This paper introduces a novel monthly video sentiment index (VNEG) by utilizing machine learning to extract sentiment from the facial expressions of millions of investors in videos. This index provides a real-time and continuous measure of sentiment. Our analysis reveals that VNEG is a robust and inversely correlated index for explaining and predicting stock returns, whether in-sample or in out-of-sample forecasting scenarios. In comparison to previous sentiment indices and commonly examined macroeconomic variables, VNEG demonstrates superior performance. Furthermore, our study highlights the notable effectiveness of VNEG in forecasting portfolios, particularly those categorized by Consumer (Cnsmr) and High Technology (HiTec) industry sectors, as well as in contexts involving large-cap stocks and high-momentum investments.

In the increasingly frequent global financial turmoil, investors prefer to invest in stable assets to hedge risks. Crude oil naturally has dual use value as a general commodity and as a financial asset, which has attracted wide attention. In this paper, we adopt a wavelet coherence analysis to study the standard of crude oil as a hedging asset and analyze the dynamic correlation of crude oil and stock market price fluctuations in the four economies of the United States, Japan, China and Hong Kong at different frequencies. The empirical evidence shows that crude oil can be conditionally used as a hedging asset for underlying securities. From the perspective of space, crude oil is suitable for investors in China’s stock market as a hedging asset, while for stock markets in the US, Japan and Hong Kong, the ability of crude oil to hedge risk has been greatly weakened. From the perspective of investment term, although crude oil cannot be regarded as a hedging asset for long-term investment, it can still play a hedging role in the short term. When the market is in a state of panic, the ability of oil to hedge risk is stronger.

We investigate all listed firms in Shanghai and Shenzhen stock Exchanges on extreme market movement days over 2010 to 2017, and highlight the important role of price limit on post extreme day stock returns. Utilising daily cash flow data of the largest trading group as a proxy of institutional investors trading behaviour, we identify institutional investors’ consistently destabilizing effects on extreme days across two markets. We further show the upper (lower) price limit hitting stocks continue to increase (decrease) for at least two subsequent days, and find evidence of long run price reversal for lower hitting stocks. Finally we find the greater net buy by large traders the higher abnormal return in three subsequent days of the upper price limit hitting regular stocks, while the net sell on extreme days tend to predict the positive subsequent abnormal returns.

This paper has applied the matrix Gaussian distribution of the likelihood function of the complete data set to reduce time complexity of multi-output relevance vector regression from O(VM³) to O(V³ +M³), where V and M are the number of output dimensions and basis functions respectively and V < M. Our experimental results demonstrate that the proposed method is more competitive and faster than the existing methods like Thayananthan et al. (2008). Its computational efficiency and accuracy can be attributed to the different model specifications of the likelihood of the data, as the existing method expresses the likelihood of the training data as the product of Gaussian distributions whereas the proposed method expresses it as the matrix Gaussian distribution.

We investigate the investment timing and financing decisions of financially constrained Small and Medium-sized Enterprises (SMEs) in a a real-option setting with asymmetric information. "Bad" firms can sell over-priced securities by mimicking in a pooling equilibrium. However, "good" firms can separate from bad firms by imposing an adverse selection cost for mimicry only when the benefit of being recognized as the "good" type outweighs the investment distortion costs. Further, asymmetric information induces good firms to accelerate investment, leading to investment distortion and higher guarantee costs. Equity-for-guarantee swap not only mitigates SMEs' financing constraints but also reduces the investment and finance distortions.