CFO gender and earnings management: evidence from China

Women can hold up half of the sky.

-Mao Tse-tung.

Abstract

We study the effect of chief financial officer (CFO) gender on earnings management (EM) in China’s listed firms from 1999 to 2011. In the cross-sectional analysis, we find that female CFO firm-years exhibit significantly lower discretionary accruals, lower total accruals, lower abnormal production costs, and higher abnormal discretionary expenditures, than the male CFO firm-years. We further examine the relation between CFO gender and EM surrounding CFO transitions. We find that the departing male CFOs are more aggressive than the departing female CFOs in managing up earnings during their last year with the firm and the newly appointed male CFOs are more aggressive than the new female CFOs in managing down earnings during their first year on the job. The evidence surrounding CFO transitions suggests that male CFOs are more aggressive than female CFOs in manipulating earnings, either in the last attempt to save their jobs or to take bigger credit for any future performance gains. Overall, our empirical evidence suggests that female CFOs engage in less EM and are more conservative in financial reporting than their male counterparts.

1 Introduction

This paper studies the relation between chief financial officer (CFO) gender and earnings management (EM) in China’s publicly traded companies. The motivations are three folds: (1) Scores of Chinese companies listed in the US have been accused of accounting frauds and the ensuing investigations by the Securities and Exchange Commission (SEC) are still ongoing; (2) Recent studies in accounting and finance show that a CFO has the most direct impact on a firm’s financial reporting and EM decisions (Mian 2001; Feng et al. 2010; Jiang et al. 2010; Ge et al. 2011); and (3) A large number of studies have shown that in general, women are more risk averse and conservative than men in making financial decisions.¹

Over the last few years, the US SEC has audited scores of Chinese firms listed in US amid concerns of accounting frauds. By the end of 2012, the SEC had delisted 50 Chinese companies from US stock exchanges and launched fraud investigations against more than 40 Chinese firms and their executives.² The implications for the parties involved are tremendous. First, this ongoing saga has gone beyond accounting standards and become an international issue between the US and China. Over the past 3 years, the SEC and the Public Company Accounting Oversight Board (PCAOB) have been negotiating, without success, with their Chinese counterpart, the Chinese Security Regulatory Commission (CSRC) in an attempt to gain access to the audit papers of the Chinese companies under investigation. The Chinese auditors argue that Chinese laws prohibit them from complying with the SEC requests. The SEC sued the accounting firms directly in the US court and won the case.³ Second, the US investors have lost billions of dollars for investing in the now delisted firms and could lose billions more if the issue is not resolved. However, the biggest losers will be the private Chinese firms seeking investor capital. Since China’s state-owned banks focus mainly on lending to large state-owned enterprises (Allen et al. 2005), small and financially constrained private firms that are the sources of most innovations are likely to suffer.

Two recent headlines capture the essence of the story: “There’s no accounting for China’s accounting” and “Chinese companies could face delisting from the US exchanges”.⁴ Clearly, earnings quality of Chinese publicly traded companies is of interest to regulators, investors, as well as capital seeking businesses. It should also be of interest to scholars.

Barua et al. (2010) examine the earnings quality in relation to CFO gender in the US firms. They find that firms with female CFOs are associated with higher quality of accruals than firms with male CFOs. They infer that female CFOs are more conservative than male CFOs in financial reporting. Our study complements Barua et al. (2010) in two aspects. First, the US and China rank the largest and the second largest economies in the world, respectively. Yet the two countries are very different in many aspects. For instance, China’s law and institution, including corporate governance, accounting standards, and quality of government are significantly less developed or weaker than those in developed countries (Allen et al. 2005). Empirical evidence shows that EM in firms located in countries with strong governance is less pronounced than EM in countries with weak governance (Leuz et al. 2003). This suggests that corporate executives such as CFOs may behave differently under different institutional environments. We examine whether the documented relation between CFO gender and EM exists in China, given the aforementioned institutional differences between the US and China.

Second, our study employs a much larger sample that consists of 11,644 firm-year observations over a longer time period (1999–2011).⁵ Our sample allows us to identify a CFO’s tenure and to observe a large number of CFO transitions which is needed when we examine the relation between CFO gender and EM around CFO turnovers. One feature of our sample worth noting is the high average fraction of female CFOs (28 %), compared to that in the US firms.⁶ In addition, as a result of Chinese government’s policies that strongly promote women’s social and economic status in the society over the last few decades, China’s corporate sector has more top female managers than many developed economies.⁷ Studies with a focus on executive gender and corporate decisions, such as this study, add another layer of relevancy in the case of China.

In the empirical analysis of EM behavior by CFO gender in the Chinese listed firms, we first conduct cross-sectional analysis on the difference in EM between female CFO firm-years and male CFO firm-years. We find that female CFO firm-years exhibit significantly lower discretionary accruals, lower total accruals, lower abnormal production costs, and higher discretionary expenses than male CFO firm-years.

We then examine the relation between CFO gender and EM surrounding CFO transitions. We find that the departing male CFOs are more aggressive than the departing female CFOs in managing up earnings during their last year with the firm, possibly in an attempt to save their jobs or to obtain bigger severance or retirement packages. We also find that the newly appointed male CFOs are more aggressive than the newly appointed female CFOs in managing down earnings during their first year on the job, possibly in an attempt to take bigger credit for any subsequent performance improvements.

Our study contributes to the existing literature on executive gender and corporate decision making by adding timely empirical evidence from China, the world’s largest developing economy. With findings from the two biggest economies in the world, one with highly developed and the other with severely underdeveloped institutions and governance, one can argue that female CFOs are more conservative in making financial reporting decisions, regardless the institutional and business environments within which they operate.

The rest of the paper is organized as follows. Section 2 focuses on pertinent research and hypotheses development. Section 3 describes the sample and EM measures. While Sect. 4 presents empirical results of the cross-sectional analysis, Sect. 5 examines gender and EM surrounding CFO transitions. Section 6 concludes the paper.

2 Related literature and hypothesis development

In this section, we start with a brief review of prior research related to gender, risk preference, corporate finance and accounting conservatism. We then discuss EM and develop a set of hypotheses related to CFO gender and EM.

2.1 Gender, risk preference, corporate finance and accounting conservatism

It is widely accepted in the literature that men and women have different risk perceptions due to the inherent genetic difference and that women are more risk averse than men. A large volume of literature further shows that gender affects personal investment choices that reflect risk preference. For instance, Hinz et al. (1997) analyze data from a 1990 Survey of Participants of the Federal Thrift Savings Plan and find that women invest their pension contributions more conservatively than men. Bajtelsmit and van Derhei (1997) study individual pension asset allocations and find that women are more likely to invest in fixed-income securities than men. Similarly, Sunden and Surette (1998) examine asset allocations of defined contribution plans using data from the Survey of Consumer Finances. They find that females and married couples are more conservative in allocation of risky assets. In a more recent study, Cesarini et al. (2010) match individual portfolio selection decisions in Sweden with the Swedish Twin Registry and find that approximately 25 % of individual variation in portfolio risk is due to genetic variation.

A few recent studies focus on gender and corporate finance. For instance, Adams and Ferreira (2009) examine how board gender diversity affects firm performance and corporate governance. They find that adding more female directors to the board has a negative effect on performance and that chief executive officer (CEO) turnover is more sensitive to stock performance if the board is more gender-diverse. On the contrary, Liu et al. (2014) show that firms with more females on boards perform better than firms with fewer females on boards in the Chinese listed firms. Huang and Kisgen (2013) examine the impact of gender of top executives (CEOs and CFOs) on corporate decisions surrounding top executive transitions in the U.S. They find that relative to their male counterparts, female top executives issue less debt and engage in fewer mergers and acquisitions. Their findings suggest that female top executives are more conservative and less overconfident in making significant corporate decisions.

There is a growing number of studies that focus specifically on financial reporting decisions and accounting conservatism (Basu 1997; Ahmed et al. 2002; Watts 2003a, b; Ahmed and Duellman 2007, among others). However, very few examine the gender effect on accounting conservatism (Francis et al. 2014). Accounting conservatism is a financial reporting approach that many firms use to limit the amount of risk in their accounting information and to avoid misleading stakeholders regarding the firm’s financial health. It is often viewed as requiring higher verification standards for recognizing a gain than a loss (Basu 1997; Watts 2003a, b). This cautious approach anticipates lower profits along with higher losses. As a result, it can lead to delayed gain recognition and accelerated loss recognition. Hence more risk-averse executives (such as female CFOs) are likely to use more conservative accounting practices than overconfident or less risk-averse executives (such as male CFOs). For example, Li and Hung (2013) find that overconfident mangers are more likely to engage in earning management behaviors. By examining EM around the CFO transitions, Francis et al. (2014) shows that female CFOs tend to adopt more conservative financial reporting policies. Barua et al. (2010) find that female CFOs are related to higher quality of accruals than their male counterparts. Their interpretation is that female CFOs are more cautious and less aggressive in making judgments related to discretionary accruals. The findings in the aforementioned studies suggest that gender of top executives has a significant impact on a firm’s financial reporting decisions as well as on firm performance.

2.2 Earnings management and hypotheses

2.2.1 Hypothesis on accruals-based earnings management

Earnings management can occur through two channels: accruals-based management and real activities-based management. Healy and Wahlen (1999) define (accruals-based) EM as follows:

“Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers.” (p. 368)

There is an extensive literature in accounting and finance examining managers’ incentives and actions to manage earnings in the United Sates,⁸ though only few focus on EM in listed firms in China. For instance, Aharony et al. (2000) examines the earnings patterns of initial public offering (IPO) firms in China and find that Chinese state-owned firms engage in “financial packaging” 2 years before selling shares to foreign investors. Chen and Yuan (2004) and Yu et al. (2006) find that China’s seasoned equity offering (SEO) firms engage in EM in order to meet the regulatory thresholds for SEO. More recent studies examine whether management demography, cross-listing, or media exposure influences accruals-based EM behaviors and whether EM affects the IPO anomalies (Cheng and Leung 2012; Eng and Lin 2012; Qi et al. 2014; Shen et al. 2014).

However, accruals-based earnings manipulations are likely to draw auditors’ attention or regulatory scrutiny, and entail a risk to managers who engage in those activities (Roychowdhury 2006). Given that corporate decisions also reflect mangers’ personal risk preferences, one would expect that the accounting practices such as financial reporting of female executives be more conservative. We hence expect that female CFOs engage less in accruals-based earnings manipulation. To capture accruals-based EM we use total accruals as well as discretionary current accruals (DCAC) estimated from the modified cross-section Jones model (Jones 1991) as described in Dechow et al. (1995).⁹ We present our first hypothesis (in alternative form) as follows:

H1

Female CFO firm-years exhibit lower discretionary current accruals and lower total accruals than male CFO firm-years do, ceteris paribus.

2.2.2 Hypotheses on real activities-based earnings management

More recently, researchers have started to pay more attention to how firms manage earnings through real activities manipulations.¹⁰ While accruals-based manipulations generally do not involve altering operations and bear no direct cash flow consequences, real activities-based manipulations affect operating cash flow. Roychowdhury (2006) defines real activities manipulation as

“departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations.” (p. 337)

Roychowdhury (2006) examines three real activities manipulation methods that can be used to avoid reporting losses: sales manipulation, reduction in discretionary expenditures and overproduction.¹¹ Discretionary expenses, such as advertising expense, maintenance, R&D, and SG&A expenses, are generally expensed in the same period in which they occur. Hence firms can report higher current period earnings by intentionally reducing such expenses. Roychowdhury (2006) argues that if the management intentionally manages current earnings upward, discretionary expenditures would be lower than normal level of expenditures, which could also lead to higher current period cash flows.

On the other hand, managers can increase production more than necessary in order to increase earnings. As argued in Roychowdhury (2006) higher levels of production help spread fixed overhead costs over a larger number of units, thus reducing fixed cost per unit. As long as the reduction in fixed costs per unit is not offset by any increase in marginal cost per unit, total cost per unit declines, which leads to lower cost of goods sold and increased operating margins. However, overproduction would cause unusually high current period production costs relative to sales, leading to lower cash flows from operations.

The net effect of real activities manipulation on a firm’s operating cash flow may be ambiguous. However, if a firm’s management intends to manage up current period earnings through real activities manipulations, its discretionary expenditures should be unusually low and its production costs unusually high. To capture real EM, we estimate abnormal production costs and abnormal discretionary expenses as in Roychowdhury (2006).¹²

One may argue that the chief operating officer is more directly responsible for production and sales decisions. However, the responsibilities of a CFO in a contemporary corporation extend beyond financial reporting and oversight. As pointed out in Mian (2001), a CFO has significant influence on major corporate decisions, including working capital management, capital budgeting and capital structure decisions. If female CFOs are more conservative in making operational decisions, one would observe less real EM activities among female CFOs than among male CFOs. We hence propose the real activities-based EM hypotheses as follows:

H2a

Female CFO firm-years exhibit lower abnormal production costs than male CFO firm-years do, ceteris paribus.

H2b

Female CFO firm-years exhibit higher abnormal discretionary expenditures than male CFO firm-years do, ceteris paribus.

2.2.3 Earnings management hypotheses surrounding CFO transitions

Motivated by Geiger and North (2006) who examine accruals management surrounding CFO turnovers in the US listed firms, we examine the relation between CFO gender and EM surrounding CFO transitions in the Chinese listed firms. Geiger and North (2006) find that the departing CFOs manage up earnings in year (t − 1), where t is the year of CFO transition, in an attempt to save their jobs or to get a better retirement package. They also find that the newly appointed CFOs intentionally manage down earnings in year (t + 1) in order to take bigger credit for any performance improvement in the subsequent years.

We build on Geiger and North (2006) by adding CFO gender into the equation. We examine the relation between CFO gender and EM in year (t − 1) and in year (t + 1). Based on our prior discussion on gender, risk preference and EM, we argue that if the departing CFOs are male, earnings manipulation in year (t − 1) would be more pronounced than in cases where the departing CFOs are female. Similarly, if the newly appointed CFOs are male, earnings manipulation in year (t + 1) would also be more pronounced than in cases where the newly appointed CFOs are female. We employ DCAC and abnormal discretionary expenditures in our examination of EM surrounding CFO transitions because manipulations of discretionary items are relatively easy and more likely to produce immediate results (Geiger and North 2006). Table 1 illustrates the expected EM by the departing CFOs at (t − 1) in Panel A and by the succeeding CFOs at (t + 1) in Panel B.

Table 1 CFO gender and earnings management by departing and succeeding CFOs

More specifically, the departing CFOs manage up earnings at (t − 1). As shown in Panel A, a CFO can manage up earnings through increasing discretionary accruals (accruals ↑) or reducing discretionary expenditures (expenses ↓). We conjecture that the departing male CFOs are more aggressive than the departing female CFOs in manipulating up earnings prior to their departures. Therefore, our focus is on column (3) in Panel A. We hypothesize that the difference in EM between the departing male and female CFOs, ΔEM, is positive measured by DCAC and negative measured by abnormal discretionary expenditures.

H3a

At (t − 1), the male departing CFOs exhibit higher discretionary current accruals than the female departing CFOs, ceteris paribus.

H3b

At (t − 1), male departing CFOs exhibit lower abnormal discretionary expenditures than the female departing CFOs, ceteris paribus.

On the other hand, the succeeding CFOs manage down earnings at (t + 1). As shown in Panel B of Table 1, a new CFO can manage down earnings through reducing discretionary accruals (accruals ↓) or increasing discretionary expenditures (expenses ↑). We conjecture that the new male CFOs are more aggressive than the new female CFOs in intentionally lowering earnings in their first year on the job. Therefore, our focus is on column (3) in Panel B. We hypothesize that the difference in EM between newly appointed male and female CFOs, ΔEM, is negative measured by DCAC and positive measured by abnormal discretionary expenditures.

H3c

At (t + 1), the newly appointed male CFOs exhibit lower discretionary current accruals than the newly appointed female CFOs, ceteris paribus.

H3d

At (t + 1), the newly appointed male CFOs exhibit higher abnormal discretionary expenditures than the newly appointed female CFOs, ceteris paribus.

3 Data and sample

We obtain the financial data from the Chinese Securities Market and Accounting Research (CSMAR) databases. The characteristic data on top corporate officers, board of directors and ownership structures is from CSMAR’s Corporate Governance Research database. We restrict our sample to all non-financial firms listed in the Shanghai Stock Exchange or the Shenzhen Stock Exchange for the period 1999–2011. We exclude firm-years with negative book equity or without the required data to calculate EM measures. Our final sample consists of 3,205 female CFO firm-years and 8,439 male CFO firm-years, with a total of 11,644 firm-year observations.¹³

Table 2 presents the numbers of female and male CFOs by year in Panel A and by industry in Panel B. The descriptive statistics in Panel A indicate that female CFO representation in China’s corporate sector is remarkably high and stable, ranging from 26 to 30 % over the sample period. Panel B shows there is a wide variation in female CFO percentage across industries. Industry classifications are based on the CRSC two-digit classifications, similar to the two-digit SIC codes in the US. A minimum of 15 observations for each industry-year grouping is required. The retail industry has the highest female CFO representation at 42 %, while tourism and culture industry has the lowest female CFO representation at 16 %.

Table 2 Descriptive statistics of CFO gender by year and by industry

Table 3 presents summary statistics for the firm, ownership, and corporate governance characteristics used in this study. Appendix 2 provides a detailed description for these variables. Firm characteristic measures are winsorized at the bottom and the top one percentile. The mean and median of variables for female CFO firm-years and male CFO firm-years are presented in columns (1) and (2), respectively. Column (3) reports test results on the mean and median differences of these variables. Several observations are noted. First, female CFOs are associated with firms that are smaller on average, as measured by market value of equity (MVE), sales, and total assets (TA). Second, firms with female CFOs tend to perform better based on return on sales (ROS) and return on assets (ROA) than firms with male CFOs do. Further, firms with female CFOs are associated with lower state ownership (STATE) and higher institutional ownership (INST), and tend to have smaller board (BSIZE) and more female directors on board (f-DIRECTOR). Lastly, female CFOs are about 2 years older than male CFOs on average. However, both female and male’s tenures on the CFO position (CFO_TENURE) are about the same, which is 3 years on average.

Table 3 Summary statistics and univariate tests of control variables

4 Empirical results

4.1 Univariate comparison analysis: CFO gender and earnings management

Table 4 presents test results based on a simple cross-sectional comparison of the means and medians of EM measures between the female CFO firm-years and male CFO firm-years subsamples. The results in Panel A show that female CFO firm-years have significantly lower DCAC and lower total accruals (TAC) than male CFO firm-years do over the whole sample period. Specifically, column (3) shows that the means of DCAC and TAC for female CFO firm-years are 1.31 % (t-statistic = 3.10) and 0.61 % (t-statistic = 2.83) lower than those for male CFO firm-years, respectively. Based on median values, the DCAC and TAC for female CFO firm-years are 0.87 % (z-statistic = 1.97) and 0.42 % (z-statistic = 2.34) lower than those for male CFO firm-years do, respectively. These preliminary results support our first hypothesis (H1).

Table 4 Mean and median of earnings management measures by CFO gender

Panel A further shows that the mean and median differences of the two real EM measures between the female CFO and male CFO firm-year subsamples are significant with expected signs. Specifically, the female CFO firm-years have, on average, 1.21 % (t-statistic = 2.76) lower in abnormal production costs (ABCOST) and 0.45 % (t-statistic = 2.11) higher in abnormal discretionary expenses (ABDXPN) than the male CFO firm-years. The median differences in ABCOST and ABDXPN show a similar pattern, suggesting that female CFOs engage less in real activities-based EM. These preliminary results tend to support our second set of hypotheses (H2).

Over the years, Chinese regulators have made progress in mandating Chinese firms to report financial statements that are consistent with International Financial Reporting Standards (IFRS). On February 15, 2006, Chinese regulators announced the issuance of Accounting Standards for Business Enterprises (ASBEs) that are largely consistent with IFRS. The mandatory adoption of ASBEs for all publicly listed firms in China began in January 1, 2007. Since the IFRS are believed to promote information transparency and good corporate governance, we examine whether the adoption of the IFRS-consistent ASBEs affects the relation between CFO gender and EM in China. We do so by splitting the whole sample into two subsample periods: the pre-IFRS period (1999–2006) and the post-IFRS period (2007–2011).

Panel B of Table 4 presents test results on the four EM measures over 1999–2006 period and Panel C for the period of 2007–2011. The results in Panel B suggest that based on both mean and median values of all but abnormal discretionary expenses measure of EM, female CFOs engage in less earning management than their male counter parts in the pre-IFRS period (i.e. supporting H1 and H2). The results in Panel C show a similar, though weaker pattern (based on median values) for the post-IFRS adoption period. Overall, our results in Panels B and C of Table 4 do not show strong evidence that implementation of IFRS affects our findings over the whole sample period.¹⁴

4.2 Multivariate regression analysis

4.2.1 Regression method

We next perform a cross-sectional multivariate regression analysis to control for other factors that have shown to influence EM. The generic regression model is as follows:

$$EM \, = \, \beta_{0} + \, \beta_{1} F\_CFO \, + \, \sum \gamma_{j} X_{j} + \, \varepsilon$$

(1)

where EM is one of the four EM measures described earlier. F_CFO is a gender indicator that equals one if the CFO is female, and zero otherwise. X_j (j = 1, 2… k) is a set of k control variables discussed below. The regression Eq. (1) is estimated by ordinary least squares (OLS) method controlling both industry and year effects. If hypotheses H1 and H2a are true, then coefficient of F_CFO, β ₁ , would be significantly negative. If H2b is true, β ₁ would be significantly positive.

As in Gerger and North (2006) and Barua et al. (2010), the set of control variables includes firm size (log of market value of equity, LOGMVE), book-to-market ratio (BM), sales growth (GROWTH), leverage (LEV), cash flow from operations scaled by lagged assets (CFFO), and lagged return on sales (ROS _t−1). The existing empirical evidence in literature suggests that firm size, book-to-market ratio, leverage, and cash flow from operations be negatively and sales growth be positively related to accruals (Ashbaugh et al. 2003; Butler et al. 2004). The lagged ROS is included to control for prior performance.

We further control for ownership structure (STATE and INST) due to its uniqueness to the Chinese listed firms, where STATE represents the percentage of shares held by central and/or local government(s) and INST represents the percentage of shares held by non-state institutions. We control for corporate governance as well since good corporate governance has shown to reduce earnings manipulation. In particular, the fraction of independent directors on the board (i-DIRECTOR) and board size (BSIZE) are found to be negatively related to discretionary accruals (Klein 2002; Xie et al. 2003). While some researchers argue that a CEO serving as board chairman of the same firm (Duality) is a good corporate governance practice, others argue for precisely the opposite. We leave it as an empirical question. We also include CFO age (CFOAGE) and CFO tenure (CFO_TENURE) to proxy for experience since CFO financial expertise is shown to influence EM (Aier et al. 2005). CFO age is also used as a proxy for risk-aversion since a person tends to become more risk averse as she/he ages. Both CFO age and CFO tenure are human factors in addition to gender that may influence corporate decision making. Finally, we control for the fraction of female directors on the board (f-DIRECTOR) and gender of the CEO (f-CEO). If women are more risk averse hence making more conservative accounting decisions than their male counterparts, more women on the board or the CEO being a woman may also influence EM of their firms.

4.2.2 CFO gender and accruals-based earnings management

Table 5 presents the multivariate regression results for the accruals-based EM measures. The dependent variable is DCAC for regressions in columns (1) to (3). Column (1) employs independent variables similar to those in Geiger and North (2006), while column (2) additionally controls for ownership structure as well as corporate governance characteristics. The main variable of interest, the gender indicator F_CFO, has a negative coefficient that is significant at 1 % level in columns (1) and (2), suggesting that the female CFO firm-years have significantly lower DCAC than the male CFO firm-years do. The results in column (2) further show that both BSIZE and CFOAGE have a significantly negative coefficient, indicating that smaller boards and older CFOs may be able to monitor firm financial reporting more effectively, resulting in less opportunities for accruals manipulation. However, CFO tenure seems not to influence the accrual-based EM.

Table 5 Multivariate regression results of CFO gender and accruals-based earnings management

As a robustness check, column (3) of Table 5 presents regression results of a performance-matched subsample. As discussed in Dechow et al. (1995), Ashbaugh et al. (2003), and Kothari et al. (2005), the modified Jones’ model may be mis-specified when applied to firms with significantly different performance, and performance matching is a convenient and reliable way to mitigate the potential problems related to misspecification. Though the past performance is controlled in our regression Eq. (1), we proceed to construct a performance-matched sample as a means of robustness check on our results. Following the procedure described in Kothari et al. (2005) we match each female CFO firm-year with one male CFO firm-year from the same industry that has the closest performance measured by ROS in that year. The performance-matched sample consists of 2,940 female CFO firm-years and 2,940 male CFO firm-years. As shown in column (3), the coefficient of F_CFO is significantly negative (−0.011 with p value = 0.045), consistent with the results in columns (1) and (2).

Regressions in columns (4) and (5) use total accruals (TAC) as an alternative measure for accruals-based EM. The coefficient of F_CFO in both columns is negative and significant at 1 % level, suggesting that female CFO firm-years have significantly lower total accruals than male CFO firm-years do. Overall, the results in Table 5 support our first hypothesis (H1) that female CFO firm-years exhibit lower discretionary accruals and lower total accruals than male CFO firm-years do.

4.2.3 CFO gender and real activities-based earnings management

Table 6 presents multivariate regression results for the real EM related hypotheses H2a and H2b. The dependent variable is abnormal production costs (ABCOST) in columns (1) to (3) and abnormal discretionary expenditures (ABDXPN) in columns (4) and (5). As in Roychowdhurry (2006) we include in the regression only firm size, book-to-market ratio, leverage and past performance to control for firm characteristics, along with ownership structure and board characteristic variables. Column (1) shows that the coefficient of gender indicator F_CFO is significantly negative (-0.011 with p value = 0.009), indicating that female CFOs engage less in overproduction as an EM strategy than male CFOs do. Column (4) shows that the coefficient of F_CFO is significantly positive (0.004 with p value = 0.060), suggesting that female CFOs engage less in income-increasing discretionary expenditures manipulation compared to male CFOs. These results are robust after controlling for ownership structure and corporate governance characteristics, as shown in columns (2) and (5) in Table 6.

Table 6 Multivariate regression results of CFO gender and real earnings management

Column (3) provides a robustness check on results for hypothesis H2a by focusing on a sample of manufacturing firms. Roychowdhury (2006) argues that manufacturing firms afford more opportunities (i.e. price discounts, overproduction, etc.) for managers to manage earnings. He finds that overproduction and sales manipulation is significantly more pronounced in manufacturing industries than in other industries. We examine if overproduction by firms with male CFOs is more pronounced than by firms with female CFOs in manufacturing industries, ceteris paribus. Consistent with findings from the full sample the finding in column (3) suggests the same EM pattern exists among firms in manufacturing industries. Specifically, we find that the coefficient of F_CFO is highly significant and negative, with a value of −0.021 (p value = 0.001). Overall, the results in Table 6 support our second set of hypotheses (H2a and H2b) that male CFOs engage more in real EM activities through overproduction and discretionary expenditures manipulation than their female counterparts.

5 Earnings management surrounding CFO transitions

5.1 Univariate comparison of variable characteristics and earnings management

Our large sample over a long period of time affords us the opportunity to examine the EM surrounding CFO transitions. Table 7 reports characteristics of relevant variables for firms with CFO transitions. Panel A shows that 205 female CFOs and 611 male CFOs departed from the CFO positions during the sample period. The CFO turnover frequency is about 6.4 % (= 205/3,205) for female CFO firm-years subsample and 7.2 % (= 611/8,439) for male CFO firm-years subsample. It also shows there is no significant difference in firm characteristics except for the firm market equity (MVE) between firms with female CFOs and firms with male CFOs. Panel B shows that among the newly appointed CFOs, 215 are females and 601 are males. On average the newly hired female CFOs are one year older than the newly hired male CFOs. Firms with newly hired male CFOs have higher sales and sales growth, and are larger in terms of TA and market equity.

Table 7 Summary statistics and univariate tests of control variables surrounding CFO turnover

The univariate comparison on EM surrounding CFO transitions concerning the hypotheses H3a, H3b, H3c, and H3d are presented in Table 8. Panel A shows that at (t − 1), the year prior to the CFO transition, ΔEM by DCAC is 3.47 % of lagged book assets and significant at the 5 % level, while ΔEM by ABDXPN is −1.62 % of lagged book assets and significant at the 10 % level. This indicates that prior to leaving the CFO position male CFOs are significantly more aggressive than female CFOs in managing up earnings by increasing accruals and/or reducing discretionary expenditures. These findings are consistent with our hypotheses H3a and H3b.

Table 8 Earnings management by departing and succeeding CFOs

Panel B shows that at time (t + 1), the first year after the CFO transition, ΔEM is −3.50 and 2.47 % of lagged book assets, measured by DCAC and by ABDXPN, respectively. Both changes are significant at the 5 % level. This indicates that during their first year on the job, male new CFOs are significantly more aggressive than female new CFOs in managing down earnings by intentionally lowering accruals and/or increasing discretionary expenditures. These findings are consistent with our hypotheses H3c and H3d.

5.2 Multivariate regression analysis

We further conduct multivariate regression to control for the relevant firm and governance characteristics. The results are presented in Table 9. Columns (1) and (2) present EM pattern by the departing CFOs at (t − 1) using DCAC and ABDXPN as dependent variable, respectively. Column (1) shows that the coefficient of female CFO indicator, F_CFO, is significantly negative (−0.043 with p value = 0.021). This result indicates that the departing male CFOs exhibit significantly higher discretionary accruals than the departing female CFOs, which is consistent with H3a. Column (2) shows that the coefficient of F_CFO is significantly positive (0.021 with p value = 0.033), indicating that the departing male CFOs exhibit significantly lower discretionary expenditures than the departing female CFOs, which is consistent with H3b. In summary, results in columns (1) and (2) show that the departing male CFOs are more aggressive than the departing female CFOs in managing up earnings at time (t − 1), the year prior to their departure.

Table 9 Regression results of earnings management by departing and succeeding CFOs

Columns (3) and (4) present EM pattern by the succeeding CFOs at (t + 1) using DCAC and ABDXPN as dependent variable, respectively. The coefficient of female CFO indicator, F_CFO, is significant at the 5 % level in both columns with the expected sign. Specifically, column (3) shows that the newly appointed male CFOs exhibit significantly lower discretionary accruals than the newly appointed female CFOs, consistent with H3c. Column (4) shows that the newly appointed male CFOs exhibit significantly higher discretionary expenditures than the newly appointed female CFOs, consistent with H3d. In summary, results in columns (3) and (4) show that the newly appointed male CFOs are more aggressive than the newly appointed female CFOs in managing down earnings at time (t + 1), the year following their appointment.

6 Conclusion

In this study, we examine the relation between CFO gender and the accruals-based and real activities-based EM in China’s listed firms from 1999-2011. We first conduct a cross-sectional univariate comparison and multivariate regression analyses. Our results show that female CFO firm-years exhibit significantly lower discretionary current accruals, lower total accruals, lower abnormal production costs, and higher abnormal discretionary expenses than male CFO firm-years do. Our results are consistent and robust for both the pre- and post-IFRS subperiods, suggesting that the adoption of IFRS consistent accounting standards in China seems to have little impact on the relation between CFO gender and EM.

We also examine the relation between CFO gender and EM surrounding CFO transitions. We find that the departing male CFOs are more aggressive than the departing female CFOs in managing up earnings during their last year with the firm, possibly in an attempt to save their jobs or to obtain bigger severance or retirement packages. The results also show that the newly appointed male CFOs are more aggressive than the newly appointed female CFOs in managing down earnings during their first year on the job, possibly in an attempt to take bigger credit for any subsequent performance improvements.

Our study contributes to the existing literature by providing the first empirical evidence on CFO gender and EM in listed firms in China, the largest developing country in the world. The ongoing accounting scandals and the ensuring SEC investigations of Chinese firms publicly traded in US stock exchanges make our study particularly timely.

Appendices

Appendix 1: Earnings management measures

1.1 Accruals-based measures

We employ DCAC as our primary accruals-based measure for earnings management. We estimate the following modified Jones model as in Dechow et al. (1995) to obtain the discretionary current accruals:

$$\frac{{CAC_{i,t} }}{{TA_{i,t - 1} }} = \frac{{\alpha_{1} }}{{TA_{i,t - 1} }} + \beta_{1,i} \frac{{(\varDelta S_{i,t} - \varDelta REC_{i,t} )}}{{TA_{i,t - 1} }} + \varepsilon_{i}$$

(2)

where CAC is current accruals, measured as the change in non-cash current assets less the change in current liabilities excluding the change in short-term debt. TA is total book assets. ΔREC is the change in accounts receivable. ΔS is the change in sales between current year and the year before.

The parameters in model (2) are estimated each year cross-sectionally by industry. The estimated coefficients are then used to calculate the normal or predicted current accruals. The DCAC is the difference between the actual current accruals (CAC) scaled by lagged TA and the predicted current accruals.

We also use total accruals (TAC) as an alternative measure, which is the difference between operating income before taxes and cash flows from operating activities, scaled by lagged TA.

1.2 Real activities-based measures

We follow Roychowdhurry (2006) and use abnormal production costs and abnormal discretionary expenditures as our real activities-based earning management measures, modeled in Eqs. (3) and (4) below, respectively:

$$\frac{{PCOST_{i,t} }}{{TA_{i,t - 1} }} = \alpha_{1} \left( {\frac{1}{{TA_{i,t - 1} }}} \right) + \beta_{1} \left( {\frac{{S_{i,t} }}{{TA_{i,t - 1} }}} \right) + \beta_{2} \left( {\frac{{\varDelta S_{i,t} }}{{TA_{i,t - 1} }}} \right) + \beta_{3} \left( {\frac{{\varDelta S_{i,t - 1} }}{{TA_{i,t - 1} }}} \right) + \varepsilon_{i}$$

(3)

$$\frac{{DXPN_{i,t} }}{{TA_{i,t - 1} }} = \alpha_{1} \left( {\frac{1}{{TA_{i,t - 1} }}} \right) + \beta \left( {\frac{{S_{i,t - 1} }}{{TA_{i,t - 1} }}} \right) + \varepsilon_{i}$$

(4)

where PCOST is production cost, defined as costs of goods sold (COGS) plus change in inventory. DXPN is discretionary expenditures, defined as the sum of research and development (R&D) and selling, general and administrative (SG&A) expenses.

Models (3) and (4) are also estimated each year cross-sectionally by industry. The estimated coefficients are then used to calculate the normal production cost and normal discretionary expenditures. The abnormal production costs (ABCOST) are the difference between the actual production costs scaled by lagged TA and the normal production costs estimated in model (3). The abnormal discretionary expenditures (ABDXPN) are the difference between the actual discretional expenditures scaled by lagged TA and the normal discretionary expenditures estimated in model (4).

Appendix 2

See Table 10

Table 10 Variable definitions