The Silenced Women: Can Public Stimulate Reporting of Violence Against Women?

Although violence against women (VAW) is widely prevalent and has myriad adverse effects, it is considerably under-reported. This paper examines whether public activism can stimulate disclosure of socially sensitive crimes such as rape and sexual assault. I investigate this question in a quasi-experimental setting arising out of an infamous gang-rape incident that took place in Delhi (India) on a moving bus in December 2012. The incident sparked widespread protests of an unprecedented magnitude, marking a nationwide ‘social shock’. Utilizing a difference-in-difference (DiD) strategy and exploiting regional variation in exposure to the shock, I find that the shock led to a sharp increase in reported VAW (apprrox. 27%), but no change in gender-neutral crimes such as murder, riots and robbery. Additional evidence suggests that the increase in reported VAW can be attributed to a rise in reporting rather than actual incidence. Using new incident-level micro-data - a repository of over 300,000 registered complaints that I compiled - I provide first evidence on retrospective reporting, and find a measurable decline in reporting-bias associated with cases of VAW, post-shock. These findings shed light on the role of social movements in moving the needle on disclosure norms. Further, improved reporting of VAW can in turn mitigate occurrence and its associated costs.

First Draft: January 2019. Latest Draft: March 2021

Coverage: Times of IndiaASE Podcast, Policy Implications Podcast

Supported by: Sigur Center for Asian Studies, International Center for Research on Women, Association for Social Economics 

Presented at: Economics of Crime Seminar (2021), ICRW (2020), SEA (2020), PACDEV (2020), MWIEDC (2020), DEVPEC (2019)WADES (2019), Gender Innovation Lab seminar series (2019), Sigur Summer Research Fellows’ Roundtable (2019)

Alcohol Prohibition and Crime: Unintended Consequences from Bihar (with Aaditya Dar)


While the relation between alcohol consumption and crime is well documented, less is known about whether alcohol regulation policies can effectively reduce crime. This paper investigates the effect of a recent alcohol prohibition policy - legislated in the Indian state of Bihar - on crime. The policy entailed a universal ban on all sale and consumption of alcohol and was strictly enforced statewide. Exploiting district-level variation in the baseline share of drinking population, we find that banning alcohol is strongly associated with an increase in crime, including crimes against women. We also find that the rise in crime is highest in time periods when, and in regions where, the ban was strictly enforced. Since state capacity and police supply is fixed, diverting law enforcement resources towards implementing the ban could reduce institutional bandwidth to prevent crimes. Our findings reiterate the need for prior assessment of resources before adopting "big-bang" reforms.

Presented at: NEUDC (2018), UEA (2018), SEA (2018)


Dynamics and Chronicity of Multidimensional and Monetary Poverty: Evidence from Ethiopia

(with Diana Ngo)

While the multidimensional nature of poverty is increasingly recognized, there is a continued interest in monetary poverty. Consequently, past scholarship has focused on describing interactions between the two. However, less is known whether these interactions result in practical policy differences for poverty targeting. Using Young Lives panel data from Ethiopia, we study the incidence, dynamics, and correspondence of five key types of poverty: (i) consumption poverty (ii) multidimensional poverty, (iii) health poverty, (iv) education poverty and (v) living standards poverty, in a static as well as chronic setting. We find imperfect yet meaningful overlap between these measures and show that poverty targeting can be improved by using measures defined for one poverty type, say multi-dimensional, to identify the poor as defined by another, say consumption. Importantly, we find that the overlap among all poverty measures improves in a chronic setting, yielding 18% to 32% gains in poverty targeting. We also identify household-level correlates of poverty and find modest overlap between determinants of different poverty types. Households that reside in rural areas, have high dependency ratio, and are headed by a female member are significantly more likely to be poor, across all poverty types. These findings indicate that policymakers can focus on a few key characteristics to target people who are deprived in multiple monetary or non-monetary dimensions for longer or shorter durations.

Impact of the Philippines Conditional Cash Transfer Program on Empowerment and Gender Based Violence: Understanding Mechanisms and Measurement (with Elizaveta Perova and Ervin Dervisevic)

This study evaluates the impact of a conditional cash transfer (CCT) program in the Philippines, i.e. Pantawid Pamilyang Pilipino Program (4P). Past CCT evaluations have largely focused on examining effects for women and children on immediate outcomes that are directly linked to CCT conditionalities, such as health, nutrition, and education. Using a combination of DiD strategy and regression discontinuity design, this study seeks to extend the scholarship on this topic in three important ways: (i) it investigates the effect of CCT on a broader set of outcomes such as empowerment, wellbeing, and exposure to gender-based violence (GBV), (ii) it assesses the effects on men, i.e. partners of grantees/beneficiaries, and (iii) it examines sustained impacts of CCT by tracking individuals who received transfers in their childhood and adolescence. Apart from distilling final effects, the study also aims to identify predominant mechanisms of impact. Ascertaining these channels can help in modifying CCT programs – a widely used social protection policy - to maximize impact across multiple facets of development.


This study also aims to assess the relative effectiveness of indirect questioning methods such as list experiments and audio computer-assisted interviews (ACASI), compared to the conventional method of face-to-face direct questioning (FTFDQ). Findings from this study component can provide pertinent evidence on data collection techniques that can be used to increase disclosure of GBV, and thereby improve estimates on GBV prevalence.

Presented at: NEUDC (2020)