Originally published here.

India is an outlier when it comes to women’s labor force participation. Over the pastdecades, the country experienced high growth rates, significant improvement in women’seducational attainment, and a remarkable decline in fertility rate. Nevertheless, the femalelabor force participation rate (FLFPR) has remained quite low when compared to otheremerging economies. The International Labor Organization (2019) places India at 179thamong 185 countries for women’s labor force participation.

Policymakers around the world have considered a broad range of policies to increase women’slabor force participation rates. These include policies related to maternity benefits (Baker andMilligan, 2008), child care support (Cascio et al., 2015), tax incentives(Eissa and Liebman,1996) and protection against discrimination at work (Neumark, 1993). While variants of most of these policies are followed globally, their relevance varies widely across countries.For instance, the need for greater child care support is likely to depend on the social context.Child care support might be a less responsive policy in societies where help from extendedfamilies is widely prevalent.

On the other hand, there could be factors which are equally ormore relevant to encourage women’s workforce participation. In this paper, we investigate therole played by the threat of sexual violence on women’s labor force participation, especiallyfor work that involves traveling away from home. Based on our findings, we explore a newline of policy to improve female labor supply in countries with a high incidence of crimeagainst women - reducing the implicit cost of traveling to work.

Sexual violence against women is widely documented to be a significant deterrent to women’sliberty to move freely, both in developed and developing nations. Past studies have providedsurvey-based evidence on how women modify their lifestyle choices to reduce the risk of violence. For example, Riger and Gordon (1981), in a study of a few cities in the US, findthat women are much more likely to avoid going out at night than men. The gravity of sexualviolence against women has been increasingly recognized at the international level, and UnitedNations now declares it as a major violation of women’s rights.

However, the incidence of such violence and the stigma borne by the victims of sexual violence vary widely. In India,several surveys report that women commonly experience sexual violence in public spaces. Ina survey of adolescent girls in Delhi, 92% reported having experienced some form of sexualviolence in public spaces in their lifetime (UN-Women and ICRW, 2013). Another survey,conducted in relatively smaller cities, found that 95% of women feel unsafe using public transport, and a similarly high fraction of women reported feeling unsafe while waiting forpublic transport, in the marketplace, or on the roads (Kapoor, 2019).

Other surveys reportthe perceived threat of sexual violence to be one of the foremost reasons discouraging womenfrom working. In their survey of non-working women in Delhi, Sudarshan and Bhattacharya(2009) find that safety concern is cited as the second most important reason for not working.The fear became particularly prominent after the Nirbhaya Delhi rape case of 2012, widelyreported in domestic and international media. Nearly 82% of the 2,500 women surveyedin several Indian cities after the attack reported leaving the office earlier (Thoppil, 2013).These surveys suggest that the prevalence of sexual crimes may discourage women who are considering whether to work or not.Extensive research exists on women’s choice to participate in the labor force. However, veryfew have tried to link the threat of sexual violence to women’s economic choices, particularlylabor supply decisions. To our knowledge, Mukherjee et al. (2001), Chakraborty et al. (2018)and Siddique (2021) are the only papers to study the relationship between sexual violence andwomen’s labor force participation.

However, both Mukherjee et al. (2001) and Chakrabortyet al. (2018) only establish correlations. While the former finds a positive correlation in aDelhi-based survey, the latter uses cross-sectional data from the India Human Developmentsurvey to find a negative relationship. The discrepancy in findings could be driven by thenon-causal approaches. Siddique (2021) is the closest to our work.

She uses data from tworounds of the National Sample Survey between 2009 and 2012 and links it with politicalevents data from the Global Database of Events, Language and Tone(GDELT) to study theeffect of any physical or sexual violence against women on women’s labor force participationin India. After eliminating district-specific factors and accounting for state-time effects, shefinds a significant reduction in women’s participation in areas with higher reported incidentsof violence.

We analyze the impact of sexual crimes against women on women’s labor force participation.We specifically focus on sexual violence since there is no apriori reason to believe that otherforms of physical violence, like murder, would affect men and women’s choices differently.Our analysis rests on a fixed-effects model using district-level panel data from India stretch-ing over a period of almost 10 years. We obtain employment information from four wavesof the National Sample Survey conducted between 2004-05 and 2011-12.

We combine thiswith official police records on district-level incidences of reported sexual crimes such as rape,molestation, and sexual harassment as opposed to media reports or self-reported perceptionmeasures. On the other hand, the GDELT data used by Siddique (2021) aggregates informa-tion on violence related to political events from a few prominent English dailies. Given thevast linguistic diversity of India and the relatively limited reach of English print media acrossthe wider population, police cases registered across India, which comprises both political andapolitical crimes, are more likely to be representative of crimes from all corners of India.However, as is true of all measures of reported crimes, print or perceived, registered crime datais also likely to suffer from measurement error problems due to large scale under-reporting(Iyer et al., 2012).

Hence, we conduct additional analysis exploiting potential exogenousvariation in crimes against women coming from variation in alcohol regulation policies acrossIndia. First, we provide estimates from an instrumental variables approach that exploitsstate-level variation in the minimum legal alcohol drinking age. We argue that restriction onalcohol sale and consumption is unlikely to affect women’s labor supply directly but closelyrelate to crimes against women (Luca et al., 2015).

Second, we use a complete alcohol banin the state of Gujarat to conduct a border-analysis. This approach compares the interiordistricts of Gujarat with those sharing a common border with the neighboring states Where there is no prohibition on alcohol sales.We find a robust and statistically significant deterrent effect of sexual crimes on femaleworkforce participation. A one standard deviation increase in sexual crimes per 1000 women(as per police reports) reduces the probability that a woman is employed outside her homeby 9.4%. In comparison, Siddique (2021) finds that a one standard deviation increase in localsexual assaults (as per media reports) per 1000 people reduces the probability that a womanis employed outside her home by 5.5%. This difference could be driven by the possibility thatwhile police records of sexual crimes may underestimate actual crime rates, media reportslikely suffer from a greater degree of measurement error.The productive employment of working-age women has important economic and social impli-cations. According to United Nations Economic and Social Commission for Asia and Pacific(UNESCAP), had India reached the same FLFPR as the US (86%), its GDP would have in-creased by an additional 42% (UNESCAP, 2007).

Considering the sizable implications, it isimperative to investigate the reasons behind the low participation rate. Previous researchershave attributed the declining trend in FLFPR in India to various supply-side and demand-side factors. One explanation is that employment for poorly educated women coming fromthe lower economic spectrum is typically driven by necessities rather than economic oppor-tunities. In the absence of education, opportunities outside the home are limited to sociallystigmatized low-skilled work.

Hence, a rising household income makes a convincing case forwomen to quit working – an income effect (Olsen and Mehta, 2006). Himanshu (2011) andWorld-Bank (2010) find a pattern of growth in female employment during financial distressconsistent with the income-effect hypothesis. In fact, a part of female-employment growth inIndia between 1999 and 2005 can be explained by the setback in the agricultural sector thatforced women to enter the labor market to supplement household income (Abraham, 2009).The decline post-2005 is, therefore, interpreted as a reversal of the increase that was initiallydriven by distress.

Higher income also leads to greater involvement of working-age womenin education, which explains some crowding out in FLFPR between 2005 and 2010 (Ran-garajan et al., 2011). As female education rises and the opportunities for white-collar jobsopen up, the income effect weakens and the substitution effect strengthens since there is nosocial stigma against white-collar jobs (Goldin, 1994; Olsen and Mehta, 2006; Mammen andPaxson, 2008). Hence, within highly educated women, the low FLFPR is partly attributedto the selection into higher education (Klasen and Pieters, 2015) and partly to the lack of suitable employment opportunities (Das and Desai, 2003).

Our paper contributes to thisliterature by estimating the extent to which the incidence of sexual violence against womenexplains women’s low labor supply in India. Encouraging women’s labor force participationby addressing longstanding social norms or ensuring adequate supply of jobs are challengingand require more long term policy interventions. Our findings raise a possibility of a moreimmediate policy intervention that could enable women to join the labor force. Further,policies directed at reducing crimes against women have first order implications aside fromimproving women’s labor force participation.

You can read the complete article here.