Replication package for "The Paradox of Seclusion: Regional Differences in Female Employment and Wages in Urban India"
Description
Why are urban gender wage gaps lower in northern than in southern states of India, despite greater gender equality (in non-wage dimensions) in the south? I show that this is due to greater suppression of women’s low-wage employment in the north, resulting in stronger positive selection: selection-corrected gaps that impute wages for the non-employed based on observed and unobserved characteristics are similar for both north and south. I suggest that stronger social norms in the north that stigmatize women’s wage work produce lower participation rates, particularly among less-educated, low-wage women who do not have access to white-collar jobs. These patterns of participation introduce significant selection biases in the measurement of gender wage gaps and help explain why urban gender wage gaps are lower in the north.
Files
Steps to reproduce
This folder contains the replication do files necessary to replicate the main results in "The Paradox of Seclusion: Regional Differences in Female Employment and Wages in Urban India." The information below provides the links from which the raw microdata was obtained, including a discussion of which rounds and variables were used. In some cases (e.g., DHS) it is necessary to register in order get access to the microdata. National Sample Survey: Employment Unemployment Schedule (NSS-EUS): https://microdata.gov.in/NADA/index.php/home [I combine four rounds (2004–5, 2007–8, 2009–10, 2011–12).] Periodic Labour Force Survey (PLFS): https://microdata.gov.in/NADA/index.php/home The NSS-EUS was discontinued after its last round in 2011-2012 and replaced with the PLFS. I pool annual PLFS data from 2017, 2018, and 2019 (excluding data from 2020-2022 due to pandemic-related disruptions). Indian Human Development Survey (IHDS): https://www.icpsr.umich.edu/web/pages/ I supplement the NSS (a cross-sectional dataset that cannot track individuals over time) with a smaller but also nationally representative panel household survey—the Indian Human Development Survey (IHDS) (2004–5 and 2011–12) (Desai and Vanneman 2018). World Values Survey (WVS): https://www.worldvaluessurvey.org/WVSContents.jsp The WVS is a cross-national survey measuring individuals’ beliefs/values on topics ranging from gender roles to political engagement. For cross-country analyses (Figure 2), I use Wave 6 (2010–2014; N=97,220 respondents, 66 countries), measuring social norms via agreement with the statement, “When jobs are scarce, men have more right to a job than women.” India-specific analysis (Table 6) combines Waves 5 (2012) and 6 (2016), restricting to urban residents in northern (N=246) and southern states (N=514). In addition to the question on job scarcity, I include other include attitudinal questions on whether “a woman earning more than her husband causes problems,” “having a job is best for women’s independence,” or “children suffer if the mother works.” All WVS estimates utilize provided sampling weights. National Family Health Survey (NFHS): https://dhsprogram.com/data/available-datasets.cfm The NFHS is a nationally representative survey of Indian households, focusing primarily on health and family welfare (IIPS 2017). I draw on the NFHS 2015–2016, restricting the sample to ever-married women aged 15–49 in urban areas across northern and southern states (7,104 and 4,595 observations, respectively). I use questions on mobility with respect to shopping or going outside town (“Are you usually allowed to go to the following places alone?”), with binary dummy for affirmative response coded. Spousal control is coded as affirmative responses to the questions on her relationship with her husband: “he insists on knowing where she is at all times,” “he does not permit her to meet friends,” “he does not trust her with money,” and “he is jealous if she talks with other men.”