The goal of this dissertation is to improve financial statement fraud detection using a cross-functional research approach. The efficacy of financial statement fraud detection depends on the classification algorithms and the fraud predictors used and how they are combined. Essay I introduces IMF, a novel combiner method classification algorithm. The results show that IMF performs well relative to existing combiner methods over a wide range of domains. This research contributes to combiner method research and, thereby, to the broader research stream of ensemble-based classification and to classification algorithm research in general. Essay II develops three novel fraud predictors: total discretionary accruals, meeting or beating analyst forecasts and unexpected employee productivity. The results show that the three variables are significant predictors of fraud. Hence Essay II provides insights into (1) conditions under which fraud is more likely to occur (total discretionary accruals is high), (2) incentives for fraud (firms desire to meet or beat analyst forecasts), and (3) how fraud is committed and can be detected (revenue fraud detection using unexpected employee productivity). This essay contributes to confirmatory fraud predictor research, which is a sub-stream of research that focuses on developing and testing financial statement fraud predictors. Essay III compares the utility of artifacts developed in the broader research streams to which the first two essays contribute, i.e., classification algorithm and fraud predictor research in detecting financial statement fraud. The results show that logistic regression and SVM perform well, and that out of 41 variables found to be good predictors in prior fraud research, only six variables are selected by three or more classifiers: auditor turnover, Big 4 auditor, accounts receivable and the three variables introduced in Essay II. Together, the results from Essay I and Essay III show that IMF performs better than existing combiner methods in a wide range of domains and better than stacking, an ensemble-based classification algorithm, in fraud detection. The results from Essay II and Essay III show that the three predictors created in Essay II are significant predictors of fraud and, when evaluated together with 38 other predictors, provide utility to classification algorithms.

There are four main components for influencing traders' behaviors in an information market context: trader characteristics, organizational characteristics, market design, and external information. This dissertation focuses on investigating the impact of individual trader characteristics on trading behaviors. Two newly-developed constructs, highly relevant to information market contexts, were identified to increase our understanding about trading behaviors: trader's economic culture and trader independence. The theory of planned behavior is used as the theoretical basis to postulate hypotheses for empirical testing. Data collected from subjects through a series of web-based experiments shows that trader participation can be fostered through recruiting individuals who are entrepreneurial, risk takers, and not highly independent traders. Additionally, a set of objective measures were developed to operationalize trader participation and performance (accuracy of prediction, and profitability). The research investigation on these concepts suggests that there is statistical evidence for a positive influence of trader participation on trader performance. In comparing the quantity influence (trader participation) and the quality influence (trader accuracy) on trader profitability, we have found that trader accuracy had higher and more significant impact than trader participation.

Market Liquidity, Active Investment, and Markets for Information. This paper studies a financial market in which investors choose among investment strategies that exploit information about different fundamentals. On the one hand, the presence of other informed investors generates illiquidity. On the other hand, investors who use different strategies can serve as quasi-noise traders for each other, thereby also supplying each other with liquidity. Thus, investment strategies can be substitutes or complements. Such externalities in information acquisition have effects on investor herding, comovement in prices and liquidity across assets, trade volume, and the informational role of prices. They further affect the relationship between financial markets and information markets. Information market competition fosters investor diversity, whereas monopoly power promotes investor herding. Also, in order to benefit from quasi-noise trading, a financial institution may engage in both proprietary trading and information sales. Security-Voting Structure and Bidder Screening. This paper shows that non-voting shares can promote takeovers. When the bidder has private information, shareholders may refuse to tender because they suspect to sell at an ex-post unfavourable price. The ensuing friction in the sale of cash flow rights can prevent an efficient sale of control. Separating cash flow and voting rights mitigates this externality, thereby facilitating takeovers. In fact, the fraction of non-voting shares can be used to discriminate between efficient and inefficient bidders. The optimal fraction decreases with managerial ability, implying an inverse relationship between firm value and non-voting shares. As non-voting shares increase control contestability, share reunification programs entrench managers of widely held firms, whereas dual-class recapitalizations can increase shareholder wealth. Signaling in Tender Offer Games. This paper examines whether a bidder can use the terms of the tender offer to signal the post-takeover security benefits to the shareholders of a widely held target firm. As atomistic shareholders extract all the gains in security benefits, signaling equilibria are subject to a constraint that is absent from bilateral trade models. The buyer (bidder) must enjoy gains from trade that are excluded from bargaining (private benefits), but can nonetheless be relinquished and enable shareholders to draw inference about the security benefits. Restricted bids and cash-equity offers do not satisfy these requirements. Dilution, debt financing, probabilistic takeover outcomes and toeholds are all viable signals because they make bidder gains depend on the security benefits in a predictable manner. In all the signaling equilibria, lower-valued types must forgo a larger fraction of their private benefits and these signaling costs prevent some takeovers. When the bidder has additional private information about the private benefits as in the case of two-dimensional bidder types, fully revealing equilibria cease to exist. This does not hold once bidders can offer not only cash or equity but also (more) elaborate contingent claims. Offers which include options avoid inefficiencies and implement the symmetric information outcome. Goldrush Dynamics of Private Equity. This paper presents a simple dynamic model of entry and exit in a private equity market with heterogeneous private equity firms, a depletable stock of target companies, and rational learning about investment profitability. The predictions of the model match a number of stylized facts: Aggregate fund activity follows waves with endogenous transitions from boom to bust. Supply and demand in the private equity market are inelastic, and the supply comoves with investment valuations. High industry performance precedes high entry, which in turn precedes low industry performance. There are persistent differences in fund performance across private equity firms, first-time funds underperform the industry, and first-time funds raised in booms are unlikely to be succeeded by a follow-on fund. Fund performance and fund size are positively correlated across firms, but negatively correlated across consecutive funds of a private equity firm. Finally, booms can make ”too much capital chase too few deals.” Reputable Friends as Watchdogs: Social Ties and Governance. To examine how governance is affected when a designated supervisor befriends the person to be supervised, this paper embeds a delegated monitoring problem in a social structure: the supervisor and the agent are friends, and the supervisor desires to be socially recognized for having integrity. Strengthening the friendship weakens the supervisor’s monitoring incentives, forging an alliance against the principal (bonding). But the agent also grows more reluctant to put the supervisor’s perceived integrity at risk, thus becoming more aligned with the principal (bridging). If the supervisor’s desire for social recognition is strong, the principal’s preferences regarding the supervisor-agent friendship are bipolar. Weak friendship makes the supervisor monitor intensively to save face. Strong friendship leads the supervisor to abandon monitoring but the agent to behave well in order to protect the supervisor from losing face. The strength of friendship necessary for the latter outcome decreases in the supervisor’s desire for esteem; that is, image concerns leverage the bridging effect of friendship. This suggests that overlapping personal and professional ties can enhance delegated governance in cultures or contexts where social recognition is important, and provides a novel perspective on issues related to crony capitalism, corporate governance, and organizational culture. / Diss. Stockholm : Handelshögskolan, 2009 Sammanfattning jämte 5 uppsatser