Corporate fraud is a worldwide problem, especially in emerging markets (e.g., China and India). If not detected and prevented on a timely basis, corporate fraud can cause significant harm not only to stakeholders of the firms directly linked to the fraud cases but also to non-fraudulent firms that compete with fraudulent firms for investors’ scarce capital or customers. Unfortunately, the detection of fraud is difficult, especially in emerging markets due to weak firm-level corporate governance and country-level investor protection. Even if fraud is detected, it may not be disclosed to the public. Hence, an important research question in academic research is to develop effective methods to detect corporate frauds on a timely basis so that the extent of damages from such fraud cases can be either completely prevented or minimised and an economy’s resource allocation becomes more efficient. The existing literature has employed a variety of fraud detection techniques, ranging from the simple logistic regression model to more advanced models such as the bivariate probit model with partial observability or the SVM (support vector machines) method developed in computer science. The objective of this proposal is to develop new fraud detection methods appropriate for different levels of financial markets (e.g., the US versus China) using new ideas and interdisciplinary methodologies.