Gender and conservatism: a cross-countries study

Abstract

Purpose

This paper aims to investigate the relation between gender and accounting conservatism in banking industry using cross-countries study.

Design/methodology/approach

The study use cross-country data in banking industry. Sample of the study consists of 202 banks from 24 countries in the period 2016–2017.

Findings

The result of the study indicates that banks that operate in high masculine society are less conservative than banks that operate in low masculine society (feminine).

Originality/value

This research suggests that investors could consider investing in a country that has low masculinity (feminine) because it is more concerned with the protection of other society members through conservative choice as a protection from misleading decisions made based on too optimistic financial report.

Citation

Purwa, P.A.L. and Setiawan, D. (2021), "Gender and conservatism: a cross-countries study", PSU Research Review, Vol. 5 No. 2, pp. 182-194. https://doi.org/10.1108/PRR-08-2019-0028

Introduction

Banking industry is important to national and global economies and is treated differently from other industries (Quttainah et al., 2013). Banking industry is considered as the mirror of economic growth and can contribute to economic development by increasing balance sheet items and through financing. Despite its importance, banks around the world have been found to manage their earnings (Shen and Chih, 2005; Cornett et al., 2009). Several allegations of international case of accounting fraud at Enron, followed by similar allegations at WorldCom, Xerox, Royal Ahold, HealthSouth and so on, have triggered a closer examination of the topic of earnings management. Recent allegation in Indonesia is Garuda Indonesia found guilty after it had changed the company’s financial position from loss to profit. In banking industry, for example US Wells Fargo, the large American bank got caught with millions of fake accounts in an instance of employees trying to meet quotas through cross-selling. Other examples include Danske Bank that got caught with the money laundering scandal; Bank of Montreal that ran an “abusive” tax avoidance scheme and inflated its losses by $288m using foreign shell companies; and Sterling Financial that was subjected to a significant, sophisticated loan scheme. The cases above are in line with Kanagaretnam et al. (2011, 2014) that in managing their earnings, banks’ financial reporting is likely to be influenced by several factors, including differences in ownership structure, bank regulation, bank monitoring and institutional factors such as creditor rights, as well as by softer dimensions such as national culture. Additionally, differences in national culture also influence a country’s adoption of accounting systems (Gray, 1988; Khlif, 2016; Armstrong et al., 2010). The differences became apparent in the recession of 2008, which had a considerably larger effect on banks in certain countries (USA) than in others (Australia, Canada and Singapore) (Kanagaretnam et al., 2014). Given the cases, this paper tries to examine the influence of national culture on financial reporting system.

This study focuses on accounting conservatism as the country’s financial reporting system. Basu (1997) states that conservatism is viewed to recognize bad news more quickly than good news in terms of earning. On the other hand, conservatism is a verification for recognizing good news rather than bad news in financial statements (Basu, 1997; Khan and Watts, 2009). Prior studies found that the more masculine the society, the lower the accounting conservatism practice, and the more feminine the society, the higher the accounting conservatism practices (Kanagaretnam et al., 2014; Salter et al., 2013). Moreover, Gray (1988) in his model posits that culture dimension may influence the degree of conservatism (income increasing, loss recognition or income decreasing).

This study focuses on dimension of national culture identified by Hofstede (2001). First, Hofstede (2001) defines culture as a collective mind program that distinguishes members from one group or category of people from another. Hofstede (2001) divides national culture into five dimensions. Individualism and collectivism describe the relationship between individuals and collectivities that apply in certain societies. Uncertainty avoidance refers to the degree to which people tolerate uncertainty and ambiguity in one country. Power distance focuses on how people receive uneven power distribution in one country. Long-term orientation refers to the time dimension of a decision. Masculinity is about the allocation of roles in one country based on sex.

Among the five dimension, Hofstede (2001) stated that masculinity focuses on expected gender role in a given society, gender has the same meaning with culture because it dictated people how to behave in certain way in accordance to their expected role. In other words, gender is an involuntary characteristic that defines human mental programming. Gender as the dimension of the national culture is divided into two categories: masculinity and femininity. Masculinity means the social gender roles between men and women in the society are clearly defined, men are expected to be tougher and ambitious; and women are expected to be tenderer, modest and concerned with quality of life. Femininity means the social gender roles between men and women in the society are overlapping, men and women have the same expected behavior such as tough, ambitious and concerned with quality of life (Hofstede, 2001).

Given the explanation, this study primarily focuses on masculinity as the dimension of national culture, which may relate to countries financial reporting system. Masculinity and femininity as the part of national culture may influence countries financial reporting system (Zarzeski, 1996; Gray, 1988). Gray (1988) posits that the higher the masculinity, the more likely it is to rank highly in terms of accounting conservatism and secrecy. Khlif (2016) found that countries with high levels of masculinity are associated with aggressive accounting manipulations and masculinity might influence country’s accounting system. In the same vein, Kanagaretnam et al. (2011) states that high masculinity is also associated with earning management in banking industry despite being highly regulated industry. In addition, Salter et al. (2013) also found that femininity is linked with high accounting conservatism.

This study differs from Kanagaretnam et al. (2014) that solely focus on uncertainty avoidance and individualism as the national culture’s dimension, the study did not explore in detail about gender’s influence on accounting conservatism. Unlike Kanagaretnam et al. (2014), this research attempts to explore the influence of gender on accounting conservatism in banking industry.

This paper contributes to the literature in several ways. First, this paper extends prior research on the relation between gender and accounting conservatism to the banking industry. Second, this research is expected to present the evidence concerning with gender influence on accounting conservatism in banking industry from cross-countries case studies. The result can be used as consideration in terms of investment choice for investor.

Literature review

Institutional theory considers that a societal framework of norms, values and taken-for-granted assumptions has an influence on how organization is operated (Oliver, 1997). Meyer and Rowan (1977) proposed that institutional environment might affect the adoption of structure and management practice in an organization. The process of adopting structure and practice is denoted as organizational imprinting. Organizational imprinting is how organization adopts certain structure features based on the way things are done instead of rational decisions (Scott, 1987). As a consequence, particular organizational practices such as how to record uncertain events and estimates is adopted because they are socially accepted method rather than rationally decided.

The socially accepted methods refer to normative expectations which have positive and negative impact toward organizations’ behavior. DiMaggio and Powell (1983) illustrated those expectations as “iron cage” that leads to institutional isomorphism. Institutional isomorphism is defined as the process of imprinting the same practices and/or structures in response to institutional pressure. In explaining the process, DiMaggio and Powell (1983) suggested three mechanisms which are: coercive that arises from political influence and cultural expectations; mimetic which resulted from uncertainty; and normative that is associated with professionalization.

Cultural value is considered as the cause of isomorphism. In organizational settings, strand of studies have investigated how culture affects company’s reporting system. Gray and Vint (1995), Zarzeski (1996) and Jaggi and Low (2000) state that national culture is associated with firm disclosure practices. Schultz and Lopez (2001), Doupnik and Richter (2004) and Chand et al. (2012) found national culture influenced the accountants interpretation of accounting standards. Guan and Pourjalali (2010), Nabar and Boonlert-U-Thai (2007), Doupnik (2008), Gray et al. (2015), Han et al. (2010) and Kanagaretnam et al. (2011, 2014) revealed that earning management practices in firms are influenced by national culture. Khlif et al. (2015) discover that national culture has effect on social and environmental reporting. It can be concluded that accounting is a product of its environment (Armstrong et al., 2010). Therefore, culture is a substantial factor in the environment as well as differences in cultural values may have a material influence on accounting and audit practices (Khlif, 2016).

Gender as the fundamental value of cultural dimensions plays an important role in how cultural value is constructed. Eagly et al. (2000) revealed that gender encompasses difference expectations toward men and women that lead to gender roles. Gender roles created patterns in which way individual would act to conform the cultural value. Hofstede (2001) introduced masculinity and femininity as the gender role patterns that prevail in both modern and traditional society. Masculine society emphasize on the achievement (Kanagaretnam et al., 2014). Feminine society is concerned with the protection of the members of society and social issues (Hussein, 1996).

The difference in gender role pattern affects the decision-making process. The decision-making process in corporate is reflected in their reporting policy. Gray (1988) argued that cultural value might affect the development of accounting system. It was argued that masculinity as one of the cultural values might influence the level of conservatism: the more masculine the society, the lower the conservatism. Khlif (2016) reviewed the extant research of Hofstede’s cultural dimensions in accounting research from 1995 to 2015. It was found that masculinity might influence the country’s accounting system. In particular, Kanagaretnam et al. (2011) studied how culture could affect conservatism and banks risk-taking in cross-countries study. They tested uncertainty avoidance and individualism as the main cultural value and later masculinity and power distance as additional proxies. They found that societies with lower masculinity are more conservative than societies with higher masculinity. In details, they found that lower masculine societies recognize more timelier loan loss provision and loan loss allowance (LLA) as the measurement for accounting conservatism. In addition, Salter et al. (2013) analyzed the influence of masculinity on accounting conservatism in 22 countries in 1989–2006 period. They concluded that in low masculine (feminine) countries, the level of conservatism is high.

Gray’s (1988) conceptual framework demonstrated that masculinity or femininity as the dimension of national culture might influence accounting conservatism. Salter et al. (2013) found that in feminine societies, the level of accounting conservatism is higher because of their concerns about protection of the weaker members of society, therefore managers and accountants act to reduce the risk that those members may invest poorly using overly optimistic reporting. On the other hand, Kanagaretnam et al. (2014) found evidence that as masculine societies emphasize achievements and competitiveness, it might lead to lower accounting conservatism. This leads to hypothesize that:

H1.

Gender is associated with the level of accounting conservatism in banks.

Method

This paper draws on a wider study on how gender affects accounting conservatism levels in banks from cross-countries setting. Following Kanagaretnam et al. (2014) and Nichols et al. (2009), accounting conservatism will be measured using bank loan loss accounting through LLA. Bank loan loss accounting reflects banks’ credit risk management activities, which are central to their profitability and risk, and can create substantial information asymmetry between owners and managers. In addition, loan loss accounting has a material effect on banks’ earnings and balance sheet amounts and requires a substantial degree of estimation and judgment. Thus, loan loss accounting should be the best place to observe bank managers’ preferences for conservative accounting (Nichols et al., 2009). The following model is used to examine the relation between accounting conservatism and gender:

(1) LLA=α+β1 HOOFS+β2CR+β3IR+β4GROWTH+β5LNSIZE+ β6SEXRAT+β7LOANRATIO+β8ROA+β9CAPRAT+e

The detailed definitions of variables are presented below in the list of variables definition. The model includes bank-level variable and country-level variable.

This study used financial data of the banks for the 2016–2017 period obtained through each bank’s website. This study selects sample countries from the 53 countries listed in Hofstede (2001). The final sample comprises 24 countries. Table 1 reports the number of banks and the number of bank-year observations by country. Table 2 reports measures of institutional variables on each country.

Variables definition:

• HOOFS = Gender as measured by Hofstede’s (2001) country masculinity score;

• LLA = Loan loss allowance at the end of year t divided by total loans at the end of year t − 1;

• Bank level variables

• LOANRATIO = Loan ratio is measured by total loans at the end of year t divided by total asset at the end of year t − 1;

• GROWTH = Asset growth is measured by total asset at the end of year t divided by total asset at the end of year t − 1;

• LNSIZE = Natural logarithm of total asset;

• CAPRAT = Total capital as reported in annual report;

• ROA = Net income divided by average total assets

• Institutional level variables;

• SEXRAT = The ratio of male and female population in a country.

• CR = Creditor right uses Djankov et al. (2007) index, which range from 0 to 4; and

• IR = Investor protection index will be measured with World Bank Index on strength of investor protection.

Findings

Before detailing the findings, Tables 3 and 4 show the descriptive statistic of variables used in accounting conservatism test. For each variable, Tables 3 and 4 report the distribution of the variable, number of countries with available data and number of bank-year observations. In Table 3, the mean value of LLA is 2.8% of total loan. Table 4 reports descriptive statistics for the country-level variables. Gender (HOOF) has means (standard deviations) of 49.13 (15.39). In addition, Tables 5 and 6 show the correlation for variable used in accounting conservatism test.

The result shown in Table 7 indicates that in Model (1), gender has negatively significant effect on conservatism. In Model (2), bank-level control variable is included and the result is still significant. In Model (3), bank-level and country control variable is included and the result also is significant. For bank-level variables, in Models (2) and (3), GROWTH and ROA are significantly associated with conservatism. For country-level variables, in Model (3), IR is negatively associated with conservatism, while CR is positively associated and SEXRAT is not significant. Therefore, H1 is supported and it can be concluded that gender is negatively associated with conservatism. Kanagaretnam et al. (2014) also stated that low masculine societies recognize timelier LLA than societies with higher masculinity.

The result is supported by previous studies by Kanagaretnam et al. (2014) and Salter et al. (2013) that found that the level of accounting conservatism is associated by gender. The more masculine the society, the lower the level of accounting conservatism. Kanagaretnam et al. (2014) state that high masculine societies concern on achievements and competitiveness that might have resulted in less accounting conservatism. In addition, low masculine societies recognize timelier LLA (Kanagaretnam et al., 2014). In this regard, Salter et al. (2013) also stated that lower masculine (feminine) societies are more concerned with care and protection of other members of society, therefore they might lead to higher conservatism to reduce risk in decision-making based on too propitious reporting.

This study also conducts several additional tests to assess the robustness of the findings. First, in Table 8, HOOF as the measurement of gender is replaced by gender equality (IGR ) as the second measurement of gender. IGR is gender equality score index obtained from Global Gender Gap Report by World Economic Forum that measures gender equality index around the world. The Global Gender Gap Index examines the gap between men and women on a scale from 0 (disparity) to 1 (parity) across four fundamental categories: economic participation and opportunity; educational attainment; health and survival; and political empowerment. The Global Gender Gap Index provides country rankings that allow for effective comparisons across and within regions and income groups. The ranking is designed to create global awareness of the challenges posed by gender disparities, and opportunities created by reducing them. The methodology and quantitative analysis behind the ranking are intended to serve as a basis for designing effective measures to reduce gender disparities. Second, in Table 9, this study separates the high and low sex ratio countries to examine the effect of sex ratio on gender and accounting conservatism. Third, in Table 10, this study also separates the sample countries according to law origin examining the effect of legal origin on gender and accounting conservatism. The classification of legal origins is obtained from Porta et al. (1998).

As shown in Table 8, in Models (1), (2) and (3), IGR is not significantly associated with accounting conservatism. The result in Table 9 shows that in both low sex ratio and high sex countries, the relation between gender and accounting conservatism is significant. In low sex ratio countries, gender is negatively associated with accounting conservatism, while in high sex ratio countries, gender is positively associated. Table 10 shows that in English, German and French law origin countries, gender is not significantly associated with accounting conservatism.

Concluding comments

The research questions addressed in this study are whether and how gender influence bank accounting conservatism and earning management. This study addresses these questions by analyzing a sample of banks from 24 countries over the period 2016–2017. In additional tests, this study explores the effect of sex ratio and law origin on the relation between gender and earning management or accounting conservatism.

The empirical results indicate that the relation between gender and earning management is not significant. In the test of accounting conservatism, the result shows that banks in low masculine societies report earnings more conservatively than banks in high masculine societies. Additionally, the result indicates that sex ratio has an effect on the relation between gender and accounting conservatism. Lastly, this study finds that law origin has an effect on the relation between gender and earning management.

The result implies that gender has no important effect on bank earnings quality. On the other hand, the relation between gender and accounting conservatism through LLA is significant. The result implies that gender has important effects on conservative accounting choice in banks. The result indicates that in terms of investment, investor could consider to invest in country that has low masculinity (feminine) because it is more concerned with the protection of other society members through conservative choice as a protection from misleading decisions made based on too optimistic financial report.

This study is subject to several limitations. First, Hofstede’s cultural variables are measured at the country level whereas our tests are primarily based on bank-level analysis. This study still assumes that the Hofstede measures are constant over time, in particular, over two years’ sample period. Second, this study only use loan loss provision as the measurement of earning management and LLA as the measurement of accounting conservatism. Based on the limitations, the future research agenda could add the sample of the study to find the relation between gender and earnings quality. Future research could use several measurements of earning management in banks industry such as loss avoidance and loan loss provision or loan charge off for the accounting conservatism. Also, future research could analyze more detailed relation between country’s law origin with gender, earning management or accounting conservatism. Comparison studies between country’s law origins are also encouraged. The current study focuses on how gender affects conservatism using banking industry. There is an opportunity to use other dimension cultures of Hofstede (2001) such as power distance, uncertainty avoidance, individualism and long-term orientation related to conservatism.

Table 1.

Sample distribution

Country	No. of banks	Total observations	(%)
Australia	8	16	3.96
Austria	3	6	1.49
Brazil	3	4	0.99
Canada	8	16	3.96
China	10	20	4.95
Denmark	4	8	1.98
Egypt	7	14	3.47
Hong Kong	8	16	3.96
Indonesia	44	88	21.78
Japan	6	12	2.97
Kuwait	6	12	2.97
Malaysia	10	20	4.95
Namibia	1	2	0.50
The Netherlands	2	4	0.99
Nigeria	7	14	3.47
Philippines	13	26	6.44
Qatar	6	12	2.97
Singapore	3	6	1.49
South Korea	4	8	1.98
Spain	2	4	0.99
Sweden	2	6	1.49
Thailand	11	22	5.45
Turkey	8	16	3.96
USA	26	52	12.87
Total	202	404	100

Table 2.

Institutional variable

Country	IR	CR	MAS	SEXRAT	MEAN GDP (In US$ million)
Australia	5.7	3	61	99.28	1,265,730
Austria	6.3	3	79	96.13	403,698
Brazil	6.5	1	49	96.67	1,924,747
Canada	7.7	1	52	98.47	1,594,405
China	4.3	2	57	106.30	11,714,347
Denmark	6.8	3	16	98.95	315,886
Egypt	4.5	2	53	102.27	284,148
Hong Kong	5.7	4	57	85.33	331,165
Indonesia	5.3	2	46	101.41	973,898
Japan	6.3	2	95	95.47	4,910,705
Kuwait	5.7	3	53	134.88	115,519
Malaysia	7.8	3	50	106.79	305,732
Namibia	5.7	2	66	96.84	12,267
The Netherlands	5.7	3	14	98.98	801,714
Nigeria	6.8	4	46	102.71	390,198
Philippines	3.8	1	64	101.36	309,242
Qatar	4.5	2	53	302.48	159,330
Singapore	8.3	3	48	97.65	316,836
South Korea	7.3	3	39	100.17	1,472,778
Spain	6.5	2	42	96.22	1,274,288
Sweden	7.2	1	5	100.11	526,250
Thailand	6.3	2	34	95.24	433,529
Turkey	6.8	2	45	96.98	857,636
USA	6.5	1	62	97.98	19,007,540
All countries (mean)	6.25	2.29	49.42	108.69	207,089,950

Table 3.

Bank-level variables

Variables	Mean	Median	Max	Min	SD	No. of countries	No. of observation
ROA	0.009	0.010	0.042	−0.097	0.011	24	404
LOANRATIO	0.677	0.681	1.433	0.092	0.159	24	404
LNSIZE	23.948	23.938	29.02	16.930	2.370	24	404
LLA	0.028	0.018	0.720	9.040	0.043	24	404
GROWTH	0.090	0.064	1.145	−0.290	0.138	24	404
CAPRAT	0.173	0.162	0.870	0.089	0.060	24	404

Table 4.

Institutional-level variables

Variables	Mean	Median	Max	Min	SD	No. of countries	No. of observation
SEXRAT	106.600	101.31	303.7	85.134	35.080	24	404
IR	6.006	6.300	8.300	3.800	1.201	24	404
HOOFS	49.138	48.500	95.00	5.000	15.399	24	404
CR	2.087	2	4	1	0.863	24	404

Table 5.

Correlation test for variable used in accounting conservatism test

Variable	CAPRAT	GROWTH	ROA	LNSIZE	LOANRATIO
CAPRAT	1.000
GROWTH	−0.046	1.000
ROA	−0.021	0.030	1.000
LNSIZE	0.071	−0.132	0.002	1.000
LOANRATIO	0.024	0.459	0.036	−0.189	1.000

Table 6.

Correlation test for variable used in accounting conservatism test

Variable	SEXRAT	IR	CR
SEXRAT	1.000
IR	−0.277	1.000
CR	−0.002	0.356	1.000

Table 7.

Regression result for accounting conservatism test

Variables	(1) LLA	(2) LLA	(3) LLA
HOOFS	−0.0004 0.052*	−0.0003 0.0919*	−0.0005 0.030**
CAPRAT		−0.043 0.3181	−0.049 0.2623
LNSIZE		−0.003 0.013**	−0.002 0.151
LOANRATIO		−1.06E−05 0.801	−2.00E−05 0.630
GROWTH		0.026 0.030**	0.024 0.049**
ROA		−0.411 0.027**	−0.414 0.025**
IR			−0.010 0.001***
CR			0.006 0.080*
SEXRAT			−7.56E−05 0.360
N	404	404	404
Adj. R2	0.007	0.036	0.058
F-statistic	3.834	3.527	3.767
Prob (F-statistic)	0.051	0.002	0.000

Note:

Significance at the 10%, 5% and 1% levels is indicated by *, ** and ***, respectively

Table 8.

Regression result for accounting conservatism test

Variables	(1) LLA	(2) LLA	(3) LLA
IGR	−0.071 0.110	0.005 0.928	−0.064 0.1474
CAPRAT		−0.089 0.418	−0.043 0.317
LNSIZE		−0.034 0.036	−0.003 0.059*
LOANRATIO		−8.89E−06 0.950	−2.43E−05 0.557
GROWTH		0.027 0.058*	0.023 0.058*
ROA		− 0.683 0.0061***	−0.393 0.033**
IR			−0.008 0.002***
CR			0.006 0.101
SEXRAT			−8.45E−05 0.309
N	404	404	404
Adj. R2	0.004	0.075	0.052
F-statistic	2.549	6.681	3.453
Prob (F-statistic)	0.111	0.000	0.000

Note:

Significance at the 10%, 5% and 1% levels is indicated by *, ** and ***, respectively

Table 9.

Regression result for accounting conservatism test

Variables	Low sex ratio	High sex ratio
HOOF	−0.005 0.042**	0.003 0.083*
CAPRAT	−0.049 0.288	0.147 0.128
LNSIZE	−0.002 0.154	0.001 0.961
LOANRATIO	−1.96E−05 0.652	0.096 0.027**
GROWTH	0.026 0.048	−0.054 0.049**
ROA	−0.410 0.037	−0.336 0.660
IR	−0.009 0.003	0.029 0.183
CR	0.006 0.127	−0.007 0.002***
N	360	44
Adj. R2	0.055	0.983
F-statistic	3.609	84.412
Prob (F-statistic)	0.001	0.000

Notes:

Significance at the 10%, 5% and 1% levels is indicated by *, ** and ***, respectively. High sex ratio means there are more male population than female population. Meanwhile, low sex ratio means there are more female population than male population

Table 10.

Regression result for accounting conservatism test

Variables	English law origin	French law origin	German law origin
HOOF	−0.005 0.698	0.001 0.678	0.007 0.233
CAPRAT	0.005 0.889	−0.288 0.024**	0.168 0.379
LNSIZE	0.009 0.118	0.002 0.757	−0.015 0.070*
LOANRATIO	0.002 0.904	−2.38E−05 0.708	−0.066 0.411
GROWTH	−0.004 0.728	0.042 0.111	0.043 0.200
ROA	−0.105 0.555	−0.513 0.127	0.165 0.912
IR	0.016 0.717	0.010 0.606	0.002 0.913
CR	0.001 0.733	0.002 0.8903	−0.002 0.6519
SEXRAT	−0.001893 0.593	0.009 0.316	−0.010 0.626
N	162	156	26
Adj. R2	0.976	0.027	0.984
F-statistic	75.863	1.479	72.368
Prob (F-statistic)	0.000	0.161	0.001

Notes:

Significance at the 10%, 5% and 1% levels is indicated by *, ** and ***, respectively. English law-origin countries are Australia, Canada, Hong Kong, Malaysia, Nigeria, Singapore, Thailand and USA. German law-origin countries are Austria, Japan and South Korea. French law-origin countries are Brazil, Egypt, Indonesia, The Netherlands, Philippines, Spain and Turkey