Is there a gender gap in corruption?

Opinion: Ananish Chaudhuri on a new study that lends little support to the idea that more women in politics will reduce corruption and misconduct, except, perhaps, briefly.

Corruption in politics
Would corruption reduce if more women took up political office?

Opinion: The idea that having more women in parliament or other positions of power will reduce corruption is well-established, with many research papers exploring the topic over the past few decades.

According to researchers at Monash University, women are generally more honest and less tolerant of corruption. Research conducted at the Massachusetts Institute of Technology finds that Indian villagers are less likely to pay bribes if the village headship is reserved for women.

Meanwhile, female mayors in Brazil have been found to be less likely to be involved in administrative irregularities.

Other papers have shown that women are more trustworthy, more averse to risk-taking and often lack the political networks required to engage in malfeasance. And some research suggests that a higher share of women in parliament or in the bureaucracy is associated with lower corruption.

So much so that in a comprehensive survey I undertook a decade ago, I suggest that women are either more honest than men or that there are no differences.

No studies find men to be more honest.

These findings have resulted in calls, from bodies such as the World Bank, for increasing the share of women in politics to improve governance and reduce corruption.

If women are less corrupt than men, will they remain so after spending time in office?

Ananish Chaudhuri University of Auckland

Governments around the world have acted on this view of women as “political cleaners”. In 2003, Mexico’s customs service announced that its new anti-corruption force would be entirely female, and in Uganda, a majority of positions as local government treasurers are assigned to women.

More than 100 countries have enacted quotas for women in elected political office.

But two questions remain. First, even if women are less corrupt than men at a point in time, will they remain so after spending time in office?

Second, the bulk of our evidence comes from surveys of ordinary citizens, often students. What about real politicians?

To answer these questions, my collaborators (from UK, Norway and Australia) and I travelled to the state of West Bengal in India. There we surveyed 400 elected politicians from the lowest level of India’s political structure, the village councils. Our work was carried out soon after the July 2018 elections.

We looked at two sets of people, 195 of whom had been elected for the first time in 2018 and another 205 who had been elected five years ago or earlier and so had experience in elected office.

The most innovative part of our study was a die-tossing task, which is a well-validated tool to study dishonest behaviour.

We gave each participant a fair six-sided die and asked them to roll the die 30 times privately. They were then to come back and tell us how many sixes they rolled with participants being paid for each six claimed.

The money paid to the participants for the entire survey, including the die tossing task amounted to nearly NZD$40 for about 90 minutes of their time. This amount is nearly equal to the total monthly honorarium they get in their role as councillors.

Given the lack of monitoring, there is a financial incentive to over-report the number of sixes rolled since reporting a higher number implies earning more money.

There is a one-sixth probability of getting a six in any one toss. So, on average, we expect people to get five sixes in 30 die rolls.

Someone reporting many more than five sixes is likely embellishing the truth. The actual extent of misreporting is low, with people reporting about eight sixes on average.

We, of course, do not know for sure who is telling the truth and who is not. However, what we can do is compare the reports of men with women as well as those from experienced and inexperienced politicians.

We find that there is a pronounced increase in female, but not male, dishonesty over time.
Among inexperienced politicians, on average women report one fewer sixes than men. But the effect dissipates with time. Among experienced politicians, women report more sixes than men.

In fact, experienced female politicians report, on average, two more sixes than inexperienced female politicians. Experienced female politicians also report, on average, more sixes than the other three groups.

Digging deeper into the factors at play, we find strong evidence for women being more risk-averse than men when taking office, and for experienced female politicians being significantly less risk-averse than their inexperienced counterparts. Given that corruption is an inherently risky enterprise, one’s risk preferences make a difference.

We also find that these patterns are driven by the sample of politicians who start with weak political networks suggesting that engaging in dishonest behaviour is predicated upon developing networks.

Gender gaps in dishonesty result from differences in experiences and socialisation. The differences between men and women may not be so great if the lives of men and women do not differ greatly - or, inversely, in a society with significant gender gaps such as India, greater differences might be expected between men and women entering politics.

Regardless, our study of real-life politicians lends little support to the idea that women’s entry into political institutions will help clean out corruption or other malfeasance - except, perhaps, briefly.

Ananish Chaudhuri is a Professor of Experimental Economics at the University of Auckland.

This article is based on a paper: Time in Office and the Changing Gender Gap in Dishonesty: Evidence from Local Politics in India to be published in the American Journal of Political Science.

This article reflects the opinion of the authors and is not necessarily that of the University of Auckland.

First published in Newsroom.

Media adviser:

Sophie Boladeras,, 022 4600 388