Good Apples, Bad Apples: Sorting Among Chinese Companies Traded in the U.S.

Accounting manipulation is considered unethical, as it violates shareholders trust and is an unjust exercise of power by the management (Fischer and Rosenzweig 1995; Gowthorpe and Amat 2005). Shareholders need to be able to trust management on the truthfulness of the financial disclosures. Firms that betray the trust of shareholders face an extremely difficult task to restore it.

A sample of these studies and their main variables of interest include financial variables (Summers and Sweeney 1998), corporate governance (Beasley 1996; Chen and Zhou 2007), financial intermediary reputation (Chen et al. 2006; Wang et al. 2010), venture capital (Agrawal and Cooper 2010), and analyst coverage (Cumming et al. 2013).

Researchers have recognized the role of opportunity in causing unethical behavior. See for example, Murphy and Dacin (2011) and Pendse (2012).

An example of non-fraudulent companies that sent signals is E House China Holding Ltd (NYSE:EJ). From the end of March, 2011 through March, 2012, the company successively announced a $50 million stock repurchase plan, a $10 million open-market purchase plan from its chairman, and a 15-percent-share cash dividend plan.

Wang (2012) examines the relationship between listing route via reverse merger and the probability of being delisted. However, delisting is different from fraudulent behavior: not all delisted firms commit fraud, or are all fraudulent firms delisted. Only 19 % of US-listed Chinese companies in Wang’s sample (263 firms) were delisted, while 37 % of US-listed Chinese companies were accused or suspected of fraud in our sample (262 firms).

The issue here is how a non-fraud-committing firm could differentiate itself after being pooled with fraud-committing firm. It is not the same as a fraud-committing firm’s attempt to rebuild its reputation after a scandal, as discussed in Sims (2009).

Wind Database is the leading integrated service provider of financial data, information, and software in China. It provides services to more than 90 % of financial enterprises, financial research institutions, and regulatory bodies.

Among the excluded 32 companies, only 12.5 % (4 companies) are fraudulent firms, against 37 % in our final sample. So, it is unlikely that the excluded companies could cause a sample selection bias in our study.

The final sample includes 11 firms that had been delisted by the end of 2011. Among these 11 firms, 7 are fraudulent firms and 4 are non-fraudulent firms. The latter were delisted as the result of going private.

We also check the robustness of our empirical results by using a more restricted definition of fraudulent firm, i.e., firms that have been prosecuted by SEC or sued by investors for financial fraud or improper information disclosure. We find all of our results are broadly unchanged under this alternate definition. Due to the space limitations, these results are not tabulated but are available upon request.

The most frequently used websites for our searches are the official SEC website (www.​sec.​gov) and www.​shareholdersfoun​dation.​com.

For example, two articles on Seeking Alpha (www.​seekingalpha.​com), titled, “China New Borun: Born to Run or Destined to Fall?” and “How China Borun Has Exaggerated Its Financial Performance,” written by Alfred Little on October 21, 2010 and October 27, 2010, respectively, questioned China New Borun Corporation (NYSE: BORN). In these articles, Alfred Little used extensive evidences to prove his view that BORN committed fraud; specifically, it (1) inflated revenues by fabricating corn germ sales; (2) understated costs of its main raw material; (3) inflated earnings by overstating its gross margin by three times and earnings per share by four to five times in 2008 and 2009 as a result of inflating revenues and understating costs; (4) hid liabilities, as a great discrepancy exists in the amount of debt between BORN’s 2009 SAT record (tax return filings submitted to Chinese State Administration of Taxation) and its F-1 filed with the SEC (the debt number in its SAT record is far larger than that reported in its F-1 filing); and (5) illegal activities, including potentially black market sales.

We classify the fraudulent activities into 14 categories: (1) investor interest damaging transaction; (2) improper issue of securities at artificially low price; (3) bad internal control; (4) frequent auditor change or bad auditor; (5) inflated or nonexistent revenues; (6) inflated earnings; (7) related party transaction; (8) insider tunneling; (9) inflated or nonexistent assets; (10) hidden liabilities; (11) committing of crime or other illegal activities; (12) hidden expenses; (13) inflated cash; (14) issued false or misleading statements. Due to space limitation, the classification is not reported but is available upon request.

Zhang and Ke (2002) conducted a survey in 2000 via Chinese Enterprise Survey System. In the survey, they sent questionnaires to more than 15,000 managers from companies located in every province in mainland China. Over 5,000 usable responses were received. The managers surveyed were from every industry and every ownership type. The level of provincial trustworthiness is assessed from managers’ response to the question: “According to your experience, could you list the top five provinces where the enterprises are most trustworthy?” The enterprise trustworthiness of a province is the weighted average score of trustworthiness ranking given by the managers. Provinces ranked as number one are assigned a maximum score of five; those ranked as number two are assigned a score of four, and so on.

Previous studies have shown a positive correlation between corporate financial fraud and earnings management (Beneish 1997, 1999a; Lee et al. 1999; Dechow et al. 1996; Jones et al. 2008). Discretionary accrual is calculated from a cross-sectional modification of the Jones (1991) model.

To calculate MSCORE, we use the estimated parameters in Beneish et al. (2011). We find that only two of the eight variables (GMI--Gross Margin Index and LEVI—Leverage Index) in the MSCORE model are statistically different at the 10-percent level in the univariate analysis. This suggests that the model may not adapt well for predicting fraudulent behaviors among U.S.-listed Chinese companies.

To calculate FSCORE, we use the estimated parameters in Dechow et al. (2011). The result is based on a small sample with only 102 observations, since financial data in COMPUSTAT is missing for number of firms.

We obtain the marginal probability effect, 0.15, by using the Stata command “mfx” following the logit regression.

In the first stage regression, we use four instrumental variables. The first three dummy variables are the minimum listing requirements demanded by NASDAQ, AMEX, or NYSE in terms of shareholder equity; pre-tax earnings; and total assets and total revenues. They are (1) ‘Requirement-Equity’ equals 1 if shareholder equity is not less than $4 million, otherwise 0; (2) ‘Requirement-Pre-tax earnings’ equals 1 if pre-tax earnings is not less than $0.75 million in the last fiscal year or two of the last three fiscal years, otherwise 0; (3) ‘Requirement-Asset and Revenue’ equals 1 if total assets and total revenue are not less than $75 million in the last fiscal year or two of the last three fiscal years, otherwise 0. It is worth noting that violating one of the above three requirements does not necessarily make a company unfit to be listed on an exchange; a company that intends to go public only need to meet one listing standard above plus some other requirements. Since these other requirements are hard to quantify, or demand several variables that are missing among our companies, we do not construct variables for them. To include the influence of market condition on companies’ choice of the means to gain U.S. listing, we follow Yung et al. (2008) and construct the fourth instrumental variable, ‘Market Heat’. It is calculated as the four-quarter moving average of the number of IPOs per quarter to its historic average going all the way back to the 1960 s. In the first stage of regression, we find that the coefficients of two instrumental variables, ‘Requirement-Asset and Revenue’ and ‘Market Heat’, are significantly negative. Companies that satisfy the minimum requirements in total assets and total revenues prefer to gain U.S. listing through an IPO rather than a reverse merger. When the issue market is “hot,” companies are more willing to choose an IPO. Due to space limitation, the first stage results are not tabulated but are available upon request.

Insider purchase is defined as corporate officers, directors and owners of more than ten percent of any type of security using their own cash to purchase stocks of the company, through either open market operations or private placement (following the definition of Lin and Howe (1990) on insiders). Stock repurchase occurs when a firm uses the company’s funds to repurchase stocks in the open market. Going private occurs when a firm proposes to purchase all outstanding shares. Dividend increase occurs when a firm increases dividend payout. New debt is obtained when a firm increases borrowing, including increasing bank loan and issuing new bonds or notes. Better auditor occurs when a firm appoints a new external auditor that has higher reputation ranking than that of the one replaced. For the definition of auditor reputation, see “Data, Sample, and Variable Description” section. New independent director occurs when a firm adds at least one more independent outside directors. Clarification via news releases is defined as having had a simple declaration made in the press or websites by top management that the company’s accounting information is of high quality.

Some may argue that securing new debt, though it does not involve paying ready cash, could be a credible signal. In bad times, when there is a large decline in equity value or a large increase in leverage ratio, the ability to secure debt must be a positive signal to investors that these firms could convince lenders to extend new debt. Given this, we reclassify new debt issue as high-cost signal in the robustness test and find all the results are still unchanged.