Exploring the Manipulation Toolkit: The Failure of Doral Financial Corporation
Eastern Kentucky University
June 18, 2016
The 2015 bankruptcy of Doral Financial Corporation, once “the best” U.S. bank according to U.S. Banker, is the largest since April 2010. The bankruptcy follows years of management manipulation and others of efforts to recover. SEC investigations revealed several frauds related to Doral Financial Corporation’s valuation of interest only strips (IOs). However, our investigation shows that Doral management’s misconduct might have gone beyond IOs valuation to include reckless-hiring, over-investing, insiders trading, and possibly opportunistic stock splits. Further, it seems that manipulation through over-hiring and overinvesting, which was described in previous literature, is not totally random. Investigating the full range of corporate misconduct help reveal new tactics that managers possibly use to pool with good firms and might help advance our understanding of the economic impact of managerial misconduct.
Exploring The Manipulation Toolkit: The Failure Of Doral Financial Corporation – Introduction
[drizzle]There is a vast literature on the measurement and detection of earnings management and managerial misconduct. However, less is known about whether manipulation affects economic efficiency or whether it merely redistributes wealth from shareholders to insiders (Kedia and Philippon, 2009). Their results show that managers pool with good firms not only through accounting manipulation but also through over-hiring and over-investing, and that these tactics could have an economic impact. They argue that their results provide novel explanation to the periods of jobless and investment-less growth. Despite the established literature on accounting manipulation (Healy and Wahlen, 1999), understanding the full range of tools that managers use to pool with more productive firms is still incomplete. Further, accounting and finance literature does not provide clear explanation of channels through which manipulation affects economic efficiency. Unfortunately, current data suites do not provide enough information to enable probing managerial decision making process around periods of manipulation.
In this paper, we utilize the manipulation and failure of Doral Financial Corporation, henceforth, Doral, to provide new insights that could help advance our understanding of the abovementioned issues. Particularly, tools that manipulative managers possibly use in addition to accounting misstatements, and the effect that these tools might have on economic efficiency.
Following Doral Financial Corporation’s scandal in 2005, the Securities and Exchange Commission (SEC) conducted an investigation to figure out what exactly went wrong with Doral. Investigators revealed several frauds related to Doral’s interest only strips (IOs) valuation and other accounting issues during the period 2000-2004. We investigate Doral’s manipulation period and the period following the discovery of the manipulation. We investigate three interrelated areas of concern related to Doral’s manipulation and failure; first, we employ the recent corporate fraud literature to figure out fraudulent behavior that Doral’s management possibly participated in and more importantly tactics that were possibly overlooked by the academic literature. Second, we investigate Doral’s executives’ compensations around the restatement period. Finally, we investigate corporate governance and house-cleaning remedies that Doral used during the post-manipulation period and how investors reacted to such remedies.
In addition to the IOs valuation frauds revealed by the SEC, our investigation highlights the possibility that Doral’s management over-hired and over-invested (Kedia and Philippon, 2009) with the ulterior motives of participating in insider trading around opportunistic split announcements (Devos, Elliott, and Warr, 2015) which quickly require reverse stock split in near future. Although claiming causal relations between Doral’s frauds and executive compensation is not possible using our research design, it is interesting to find out that Doral started granting stock options to executives late 1999, few weeks before the start of the manipulation era. It is also interesting to find out that option grants were always timed immediately before earning announcements.
Our investigation shows that Doral’s “pooling with good firms” possibly took several forms in addition to those described in the extant literature. Doral appeared to “geographically pool” with larger financial institutions through expanding their operations to New York City. They also pool through changing the physical appearance and size of their new branches so they look more like those of larger firms.
Doral manipulation and decline provides a perfect environment to better understand the hire/fire and invest/divest tactics described in Kedia and Philippon (2009). Our investigation shows that expansion and contraction of Doral were not totally independent. Doral did not simply shed employees and capital when manipulation was detected. Instead, Doral fired the same exact employees they initially hired and shut off the same exact branches they initially launched. These non-random business moves in-and-out of specific geographical areas would affect unemployment rates on those areas and would provide a novel explanation to the variation in unemployment rate fluctuations among states. I.e., if manipulative firms frequently expand to areas at which larger/more successful firms are located and eventually get rid of such expansions, then we should observe an uneven unemployment rate fluctuations among states with high number of larger firms and states with low number of large firms. Particularly, states with high numbers of large firms are expected to have significantly higher unemployment rate fluctuations. Our results show that the average standard deviation of the unemployment rate for states with less large firms is 1.42, while it is 1.76 for states with more large firms.
Doral Financial Corporation’s post-manipulation management adopted multiple corporate governance enhancements such as chair/CEO role separation and increase board size and independence. Our announcement returns results indicate that cumulative abnormal returns (CAR) are significantly positive for governance and house-cleaning remedies. The five day CAR following role-separation announcement is 6.7%. Announcement returns of CEO and treasurer resignation and the CFO termination is 5.2%. Further, announcement returns of the resignation of president/COO is 4.6%. Investigating Doral’s remedies post-announcement price movements and analysts’ coverage shows that corporate governance remedies undertaken were not enough to restore Doral’s good pre-restatement levels. Failure to restore reputation could reflect the fact that the reputation hit that Doral received was just too strong.
This paper provides two new insights to the corporate manipulation literature. First, the idea of endogenous manipulative stock split. To the best of our knowledge, this is the first paper to claim that stock splits can be used by executives to achieve personal gains through insider trading. A similar claim is provided by Devos, Elliott, and Warr (2015), however, stock split is exogenous in their paper, so they don’t claim that split is a manipulation tactic by itself. Second, the idea that over-hiring and over-investing (that was quantitatively described by Kedia and Philippon (2009)) is not random, instead we claim that it could result in oriented expansions toward specific states/cities. Besides providing insights about tools of managerial misconduct and the possible channels through which manipulation might impact economic efficiency, deeply investigating the bankruptcy of Doral Financial Corporation has several other benefits. First and most important, the current setting enables us to test market reaction to role separation and house-cleaning announcements. This is a test that is not possible using a standard panel data setting due to lack of data availability in standard data suites. Second, this design enables us to investigate compensation plans in a manner not possible using archival type research. Annual reports provide communications regarding incentive packages design and