Corporate Fraud and Regulation: The Never-Ending Game of Cat and Mouse

From the original Ponzi scheme of 1920 to the collapse of Enron in 2001, Lehman Brothers in 2008, and Wirecard in 2020, the history of the financial markets is marred by a continuous stream of financial scandals. Billions of dollars were lost as a result of these financial disasters, which shook investors’ confidence, destroyed companies, and ruined people’s lives. With the fallout of every major financial scandal comes the public outcry for regulations and reforms.

Yet, despite the passage of many tough regulations with the aim of cracking down fraud, a new wave of corporate scandals somehow always finds a way to resurface. Could it be that fraud is a persistent feature of the financial markets that will never go extinct? And if financial reporting failure is indeed a permanent risk, then to what extent can anti-fraud regulations achieve their stated goals of cracking down on fraud? Our study, ‘Everlasting Fraud’, investigates these two questions by being the first to probe the interdependent mechanisms of corporate fraud and anti-fraud regulation.

We focus on accounting fraud and begin by building a multi-period model featuring a representative firm and a regulator. At the end of each period, the firm manager privately observes the firm’s true economic earnings but may issue a potentially biased earnings report to the market. The market does not observe the firm’s economic earnings and can only update its estimate of the firm value based on the manager’s report. The regulator utilizes a detection technology to inspect the firm’s report. With a certain probability, the technology uncovers fraud in the report and reveals it to the market. The model setup is thus a fairly accurate portrayal of the ‘game’ of fraud and detection.

The manager’s goal is to maximize firm value. In doing so, he weighs the benefit of inflating the firm’s earnings report and the cost of being caught by the regulator. The regulator’s goal is to maximize the amount of information that an average investor can gather from the firm’s report. In doing so, she weighs the benefit of uncovering fraudulent reporting and the cost of detection. A key takeaway from the model is that the manager’s and the regulator’s calculations are intertwined in that their benefits and costs both critically depend on ‘fraud-induced uncertainty,’ that is, the information uncertainty about the firm is induced by the cumulative fraud committed in the past. A higher level of fraud-induced uncertainty incentivizes the manager to commit additional fraud because he knows that the investors are eager for new information and will value even the biased report that he releases. At the same time, it also disciplines the manager because he anticipates that a higher level of past fraud will likely invite closer scrutiny from the regulator. A higher level of fraud-induced uncertainty motivates the regulator to detect fraud because it would be more valuable to clear uncertainty and restore information precision. As such, the two players’ calculations move hand-in-hand, and in equilibrium, regulation intensity matches the cumulative fraud committed by the firm.

The analysis above begins to tell why fraud may never cease to exist: if a regulator anticipates a low level of information uncertainty about the firm and thus a low payoff from detection, she would spend little on detection because trying to catch fraud consumes regulatory resources. In anticipation of low detection risk, the firm manager continues to commit fraud and thus fraud gradually builds up. As fraud reaches a critical level, the regulator would concentrate resources on the firm and the increased detection risk eventually dampens the manager’s incentive to commit fraud. Upon detection, fraud is cleared in the firm, and the cycle repeats. This rationale explains the time-series persistence of fraud within firms: fraud never ends, because as fraud drops, it becomes socially optimal to reduce costly detection, which in turn creates the hotbed for future fraud.

Expanding our analysis to a multi-firm setting, we show that, as an unintended consequence, the regulator’s allocation of resources across firms in fraud detection efforts tends to synchronize firms’ fraud decisions and eventually contributes to the emergence of fraud waves. This happens because, as firms start out with different levels of cumulative fraud, the regulator concentrates her detection resources on the most fraudulent firms, while leaving firms with the intermediate and low fraud levels under the radar. As the optimal response to the anticipated detection risk, the most fraudulent firms cut back on committing new fraud while other firms become more aggressive, which shrinks the gap between their cumulative fraud levels (ie the ‘catch-up’ effect). The convergence of fraud of different firms increases the incidence of them being detected to cluster in time and possibly gives rise to fraud waves. Ironically, after a firm is caught by the regulator and its cumulative fraud is cleared, the firm would continue playing the ‘catch-up’ game in committing fraud, because it factors in the regulator’s optimal allocation of detection resources and anticipates its fraudulent activities being masked by other uncaught, fraudulent firms. This rationale explains the cross-sectional persistence of fraud across firms.

We provide four sets of empirical results that are consistent with our model predictions. First, we document that analysts’ revision of earnings forecast for a firm is more responsive to unexpected earnings when the firm’s information uncertainty is high. This finding confirms that information uncertainty boosts the value of accounting reports and the potential return from reporting fraudulently. Second, we show that a firm is more likely to be caught for having committed fraud during periods of high information uncertainty. This finding is consistent with the regulator rationally allocating more resources towards firms with higher levels of built-up fraud. Third, we document a significant hump-shaped relation between a firm’s fraud amount and the level of information uncertainty. This hump-shaped relation reflects the two countervailing effects in our model, namely, the higher incentive to commit fraud and the higher likelihood of being detected (as the firm starts to attract regulatory attention). Last but not least, we find that fraud level indeed tends to converge across firms in the sense that firms with higher levels of information uncertainty are more cautious about committing additional fraud while firms with lower levels of uncertainty are more aggressive at committing fraud. 

Our findings, to be clear, do not suggest that anti-fraud regulations are ineffective. Quite to the contrary, anti-fraud regulations are very effective at tamping down fraud temporarily by sharply decreasing the most fraudulent firms’ net benefits from continuing fraud. However, even the toughest anti-fraud regulations cannot eradicate fraud as long as detection is costly. As such, fraud and detection become a game of cat-and-mouse that never ends, not in a ‘humans are born greedy’ kind of way but in a perfectly rational way. To conclude, fraud should be viewed as a permanent risk in the financial market and investors need to have a realistic understanding of this risk.

Vivian W. Fang is an Associate Professor of Accounting at Carlson School of Management, University of Minnesota.

Nan Li is an Assistant Professor of Accounting at Carlson School of Management, University of Minnesota.

Wenyu Wang is an Associate Professor of Finance at Kelley School of Business, Indiana University.

Gaoqing Zhang is an Associate Professor of Accounting at Carlson School of Management, University of Minnesota.