Principal investigator: Viviane Pires Ribeiro
Paper Title: Women and corruption: What positions must they hold to make a difference?
Authors: Chandan Kumar Jha and Sudipta Sarangi
Location of the Intervention: European countries
Sample Size: 155 regions from 17 European countries
Main theme: Genre
Main Variable of Interest: Corruption
Type of InterventionAnalysis of the positions women should occupy to reduce corruption.
Methodology: Instrumental variable analysis
Gender inequality persists in all countries, existing in access to education, work, and participation in economic and political activities. Given the debate that arose just over a decade ago, that women possibly behave differently from men in many economic circumstances, Jha and Sarangi (2018) examine the roles women play in impacting corruption, focusing on female participation in the workforce and their presence in parliament. The study's conclusions suggest that female participation in politics should not only be encouraged to achieve gender equality, but also because it has positive externalities – a negative impact on corruption.
Evaluation Context
Corruption remains a significant issue in both developed and developing countries because of its negative impact on economic growth and development outcomes. Just over a decade ago, the debate arose that women may behave differently from men in many economic circumstances.
In the literature addressing the impact of gender on corruption, some studies have found a negative correlation between female representation in parliament and corruption, while other studies express concerns that this observed negative association between gender and corruption was not causal and was likely motivated by the omission of other factors that may be correlated with women's participation and/or corruption in a country. In this context of intense debate, Jha and Sarangi (2018) address the concerns raised in this literature, first seeking a causal relationship between gender and corruption through instrumental variable analysis and, second, adopting a more nuanced approach to this problem, identifying the different economic roles that women can assume vis-à-vis corruption and investigating the impact of each on corruption.
Intervention Details
Given that much of the literature on corruption is affected by a lack of instruments or weak instruments, Jha and Sarangi (2018) make a methodological contribution by drawing inferences based on Moreira's (2003) conditional likelihood ratio approach, using data from 155 regions in 17 European countries.
The primary measure of corruption used in the study is the Corruption Control Index (CCI) published by the World Bank. The CCI is a continuous variable that takes values from -2,5 (most corrupt) to 2,5 (least corrupt). The authors used the negative of the CCI in all specifications so that a higher number indicates more serious corruption. The CCI was constructed so that its mean is zero and its standard deviation is equal to 1. The aim of the CCI is to capture perceptions of the extent to which public power is exercised for private gain, including small and large forms of corruption, as well as the “capture” of the state by elites and private interests.
The data for women's participation in the workforce (WP) were from the International Labour Organization (ILO). The United Nations Statistics Division (UNSD) provided data on the proportion of women in administrative positions and the proportion of women in decision-making positions. Data on the percentage of women in parliament were compiled by the Inter-Parliamentary Union (IPU) and were retrieved from the World Bank. All measures of female participation used by the authors are the percentage of women in the respective category.
The World Bank's Gross National Income per capita – formerly Gross National Product (GNP) per capita – in US dollars was used as a measure of income. The Association of Religion Data Archive (ARDA) provided data on the proportions of Christians and Muslims in the total population in 2005, the last year for which such data was available. Data on the countries' colonial history were taken from Treisman (2007). Freedom House assigns a score of 1 to 7 for political rights – a score of 1 indicates that citizens enjoy a wide range of political rights, while a rating of 7 implies few or no political rights.
Methodology Details
In the econometric model, the authors considered the corruption index in the country. i as the dependent variable. The other independent variables are: the proportion of women in different occupations in the country. i, depending on the specification; GNP per capita; political rights; proportions of Christians and Muslims in the total population; a dummy variable that takes the value 1 if the country is a former British colony and zero otherwise; and another dummy variable that takes the value 1 if the country was never colonized and zero if it has a colonial past.
GNP per capita was added as a control variable in all specifications because higher-income countries may be able to curb corruption more effectively than developing countries. The authors expect strong political and democratic institutions to have less corruption. Therefore, the “political rights” published by Freedom House were included as an additional regressor in the model. The literature has shown that cultural factors and social norms impact corruption. To capture these aspects, the authors included proportions of Christians and Muslims in the total population as additional regressors. Furthermore, recent studies have found that a country's colonial past, its origins, can affect corruption through its impact on economic and political institutions. It is argued that a colonized country inherits the institutional configuration of its colonizer, which is likely to persist after independence. In this sense, a “Former British Colony” and “Never Colonized” dummy variables were included in the model.
The authors used instrumental variable analysis, which addresses issues of omitted variable bias and potential reverse causality, as well as instruments to establish causality and make inferences based on the conditional likelihood ratio approach proposed by Moreira (2003), the Anderson-Rubin statistic (Anderson and Rubin, 1949), and the LM-J statistic (Kleibergen, 2002).
Results
Jha and Sarangi (2018) provide robust evidence that the presence of women in parliament has a causal and negative impact on corruption, while other measures of female participation in economic activities have no effect. Furthermore, this negative relationship between women's presence in government and corruption is also found in a regional analysis of 17 European countries, alleviating concerns that the relationship is driven by unobservable characteristics fixed by the country. Subsequently, the authors show that this relationship does not disappear when women achieve similar social status.
In statistical terms, the results indicate that the relationship between women's presence in politics is not only statistically significant but also considerable. Even with the lower Ordinary Least Squares coefficient (instrumental variable), an increase of one standard deviation (9,77%) in women's participation in parliament is associated with an improvement in the expected perception of corruption of 0,26 (0,45) points. This is considerable, given that the index itself is measured on a scale of -2,5 to 2,5 and the standard deviation of the index is 1,03. In the regional analysis, an increase of one standard deviation (about 10 percentage points) in women's participation in local government is associated with a significant reduction in bribery of one-tenth of the standard deviation.
Lessons in Public Policy
Jha and Sarangi (2018) point out that the term “workforce” used in previous studies is too broad a measure and does not make clear how women affect corruption. For example, women may affect corruption if they are less corrupt and accept fewer bribes than men. Alternatively, women may affect corruption when they are in positions of power, drafting and implementing strict anti-corruption laws within their organizations or better enforcing existing laws. Since female participation in the workforce consists of women in both roles, it is important to distinguish which of these roles (or a combination of both) is associated with lower corruption. To capture these roles, the authors present two additional measures of female participation in economic activities: (i) the proportion of women in administrative positions and (ii) the proportion of women as legislators and managers.
The analysis shows that the presence of women in the workforce, in administrative positions, and in senior-level decision-making roles is not significantly associated with corruption in a country. This lack of correlation is noteworthy, as it suggests that women are not inherently less corrupt. Furthermore, the results indicate that: women's participation in local government is associated with less bribery; and women have a systematic negative impact on corruption only if they are represented in parliaments, implying that the effect on corruption is possibly through policymaking.
Furthermore, it is possible to refute the speculation that the observed gender differences in corruption are motivated by gender differences in social status. In fact, the analysis suggests the opposite: corruption is lower if women enjoy greater status equality with men, possibly because they are more able to influence policymaking.
Finally, Jha and Sarangi (2018) raise the following question: how do women reduce corruption by being in politics? One possible answer, according to the authors, could be that they favor different policies than those advocated by men. Recent research has extensively explored the political implications of gender representation in government. It has been reported that women in local government in India allocated a larger share of the budget to public goods more closely associated with women's concerns, as well as to the provision of basic infrastructure needs, and were more concerned about whether subsidies were provided to the target group without corruption. Furthermore, it has been found that female political representation is also positively associated with state spending on health and educational outcomes. At the same time, there are also studies showing that education reduces corruption.
References
JHA, Chandan Kumar; SARANGI, Sudipta. Women and corruption: What positions must they hold to make a difference?. Journal of Economic Behavior & Organization, v. 151, p. 219-233, 2018.
