I am an Assistant Professor in the Department of Economics at IIT Bombay, specializing in health economics. My research focuses on innovative causal inference methods, particularly utilizing latent factor models for identification. I am passionate about addressing disparities in access to healthcare and health outcomes, especially for vulnerable populations. My work aims to provide insights that can inform policy decisions and improve health equity.

Choosing Wisely: Evaluating Latent Factor Models in the Presence of Contaminated Instrumental Variable with Differing Strength

joint with Anirban Basu and Shubham Das

December 2024

Abstract Causal inference methods are widely used in empirical research; however, there is a paucity of evidence on the properties of shared latent factor estimators in the presence of contaminated instrumental variable (IV) when strong IV may not be available. We present a theoretical formulation to depict how the strength of an IV and the degree of contamination simultaneously determine the optimal choice of estimator. We perform Monte Carlo simulations with four outcome variables and an endogenous treatment variable, with sample sizes varying between 500 and 1000, and for 1000 iterations, to compare the finite sample properties of the OLS, 2SLS, shared latent factor (SLF), and shared latent factor with IV (SLF+IV) estimators. Finally, we demonstrate the applicability of the proposed estimators to study the causal impact of obesity on different health indicators: diastolic blood pressure, systolic blood pressure, blood glucose level and hemoglobin level, using data from the 2019-21 Round 5 of the National Family Health Survey (NFHS-5) from India. Our simulation results indicate that, for a given strength of the IV, there exists a threshold degree of contamination, such that the SLF+IV estimator has a lower (greater) bias than the SLF estimator when the degree of contamination lies below (above) that threshold. Similarly, we find that for a given degree of contamination of the IV, there exists a threshold strength of the IV, such that the SLF+IV has a lower (greater) bias than the SLF estimator if the strength of the IV lies above (below) that threshold. The empirical results suggest that obesity is significantly associated with higher diastolic and systolic readings, a higher blood sugar level, and a higher haemoglobin level in the blood.

COMET-LF: A Compartmental Model of Dynamics of Infection, Disease, and Elimination Strategies for Lymphatic Filariasis

joint with Indrajit Ghosh, Suchita Nath-Sain, Shoummo Sen Gupta, Chhavi Pant Joshi, Tanu Jain, Swaminathan Subramanian, Mithun Kumar Mitra

September 2024

(Under review)

Abstract Lymphatic filariasis (LF) is a mosquito-borne neglected tropical disease (NTD) caused by filarial worms. India accounted for 55% of the global population at risk of LF in 2021. The World Health Organization (WHO) has targeted LF elimination by 2030; however, India aims to achieve LF elimination prior to the global WHO NTD target. Mathematical models are useful tools to evaluate and guide elimination strategies. We propose a new compartmental model—COmpartmental Modelling of Elimination strategies and Transmission of Lymphatic Filariasis (COMET-LF)—to assess the impact of mass drug administration (MDA) on LF elimination. Our model incorporates drug efficacy data from a clinical trial and generates estimates of disease (lymphoedema and hydrocele) prevalence. The model is calibrated to publicly available microfilaria (Mf) and disease prevalence data (2008-2013) from Bihar, India. Predictions of the number of MDA rounds needed for achieving the elimination threshold were generated for various endemic scenarios. The projected estimates were compared with established micro- (LYMFASIM) and macro- (EPIFIL) simulation models for LF transmission. Disease burden estimates and the impact of MDA on disease burden were generated using COMET-LF for different endemic scenarios. Our simulations suggest that the disease burden reduces over much longer timescales - 20 years for a reduction of 8%-11.5% following 5 rounds of MDA. We extended COMET-LF to a meta-population model to investigate the role of migration among neighbouring regions on elimination and resurgence probabilities. We found that high Mf prevalence in the spatial neighbourhood can increase the number of required MDA rounds for elimination up to 3 additional rounds for the two-drug regimen. Furthermore, we assess the impact of migration on the resurgence probability in a non-endemic region which is spatially adjacent to a high-Mf

Childbirth and Gender Disparity in Labour Market Outcomes: Evidence From an Event Study

joint with Preeti Jaiswal and Sankar Mukhopadhyay

August 2024

Abstract Despite increasing economic growth, rising educational attainment, and lowering fertility rates, there exists persistent gender divergence in labour market outcomes in almost all countries. One of the significant factors, contributing to this existing gender gap is the ‘motherhood penalty’, defined as the negative impact on various labour market outcomes such as labour force participation, earnings, and work hours resulting from the disproportionate burden of childcare falling exclusively upon the mothers. There exists scarce literature focusing on the impact of childbirth on female labour market outcomes at the extensive and intensive margins in the Indian context, and particularly its dynamic impact as a child grows up. The present study aims to estimate the causal impact of new childbirth on mothers’ labour force participation, earnings, and work hours compared to fathers and married women without children to isolate the gender-specific effect of motherhood. Using the longitudinal data from the Consumer Pyramids Household Survey (CPHS) over the period 2016 to 2023, we employ a difference-in-differences event study approach and use fathers and married women without children as control groups for the analysis. To address the self-selection of the event (childbirth), we perform kernel matching, nearest neighbour matching, and entropy balancing, which largely validate the parallel trends assumption. The likelihood of mothers’ labour force participation shows a decline of 4% at the time of childbirth in rural regions, while in urban regions, it shows an immediate decrease of almost 4% which further decreases to 10%, four years after childbirth compared to fathers. We find no significant decline in mothers’ earnings post-childbirth in rural regions, instead, earnings increase sharply by 25% a year later compared to fathers with no effect in urban regions. Moreover, relative to married women without children, we find a significant increase of 6% in the likelihood of mothers’ labour force participation two years post-childbirth in rural regions with no impact in urban regions and an increase of 33% and 18% in mothers’ earnings a year after childbirth in rural and urban regions, respectively. We do not find a significant impact on work hours. The decline in labour force participation reveals the immediate impact of childbirth, but unlike other countries, this decline is temporary and eventually becomes negligible. Our findings on the motherhood penalty, independent of the gender-specific effects, suggest that mothers prioritize their labour market potential compared to non-mothers, though this shift usually takes about two years to materialize. These findings underscore the need for policy interventions that offer affordable childcare and more flexible work arrangements to support a smoother, more timely reintegration of mothers into the workforce.

​

Caste Disparities in Awareness of Hypertension and Uncontrolled Blood Pressure in India

joint with Shoummo Sen Gupta, Swapnil Gupta and Kshitiz Udainiya

November 2024

Abstract Hypertension often remains undiagnosed, leading to unawareness of its status and adverse health outcomes such as brain stroke, cardiovascular diseases, and early mortality. Prior research shows individuals from lower economic and social background to have lower levels of health literacy and higher likelihood of being unaware of their hypertension status. Minority caste groups - scheduled caste (SC), scheduled tribe (ST), and other backward caste (OBC) in India lag behind in terms of educational attainment, health status, and access to healthcare services. This study examines caste disparities in awareness of hypertension and having uncontrolled blood pressure (conditional on being on hypertensive medication) for the minority caste groups relative to the general caste in India. Further, we explore mechanisms driving caste disparities in awareness. Data for the study comes from the Longitudinal Aging Study in India Wave 1 (2017-18) and the National Family Health Survey 2019-21 (NFHS-5). We exploit available information on self-reported hypertension and objective biomarker data on blood pressure to operationalize the definition of awareness: self-report having hypertension, given that individual has the disease. To account for unobserved factors correlated with caste and the binary choices of: (a) self-report hypertension, (b) consent for medical examination, (c) have hypertension (biomarker positive or on hypertensive medication), we estimated a trivariate probit model with selection, including the three choice probabilities. An extension of Blinder Oaxaca decomposition was used for identifying the driving factors determining the caste disparities in the awareness of hypertension. The minority caste groups have significantly lower educational attainment and higher rates of smoking and alcohol consumption. The SC, ST, and OBC are 0.5% (SE=0.00001), 7.5% (SE=0.0001) and 2.6% (SE=0.00003) less likely to self-report as having hypertension while they have the disease. The ST are 0.6% (SE=0.00001) more likely, whereas the SC and OBC are 0.8% (SE=0.00001) and 1.1% (SE=0.00002) less likely to have uncontrolled blood pressure. Increase in educational level, decrease in current smoking and alcohol use among the minority castes are likely to reduce the caste disparity in awareness of hypertension between the minority and the general caste. Policies aimed at improving the educational level of minority groups and disincentivizing smoking and drinking behaviours can reduce the gap in awareness of hypertension.

Securing tomorrow’s health: A study of the causal impact of maternal health investments on child health outcomes

joint with Titas Chowdhury

November 2024

Abstract Maternal and child healthcare remains a major challenge to the global public health system, especially in developing countries. Maternal healthcare in the form of antenatal and neonatal care determines the well-being of the child in the formative years as well as the long-term productive capacity of the child. In India, particularly, maternal health care varies largely with the socio-economic and demographic status of the mothers and their subjective health investments in their children especially in their infancy and formative years. In this study, we examine the relationship between the child’s health inputs (like prenatal care received by the mother during pregnancy and institutional delivery of the child) and child anthropometric health outcomes (like stunting and underweight) using pooled cross-sectional data from the Indian National Family Health Survey (NFHS) of 2015-16 and 2019-21. We account for the endogeneity of child health inputs using an instrumental variable approach taking the frequency of watching television by the mother as an instrument. The anthropometric measurements of the child are available only for a subset of the total sample, giving rise to a potentially self-selected sample, as the sample of mothers who agree to the biomarker schedule and who do not are intrinsically different across observable characteristics, and therefore assumed to be different across unobservable characteristics as well. We account for the non-random selection using the Sartori (2003) estimator, which does not rely on the exclusion restriction in the two-step Heckman method for addressing self-selection and works well in our context. We also explore some of the potential mechanisms like the “Safe Motherhood” intervention program - “Janani Suraksha Yojana” launched by the Government of India in 2005 to propagate institutional delivery among poor pregnant women through which the child health inputs can affect child health outcomes, particularly in the economically backward states of India. Accounting for the endogeneity of health inputs, we find mothers in rural areas who opt for prenatal care during their pregnancy to be 55.8% (NFHS-5) and 36.9% (NFHS-4) less likely to have children who are stunted, relative to mothers who do not opt for the recommended number of prenatal care visits. Similarly, mothers in rural areas are 46.8% (NFHS-5) and 23.3% (NFHS-4) less likely to have children who are underweight if they opt for the recommended number of prenatal care visits. Exploring the potential mechanisms, we find that when mothers from the most marginalized section of the country, namely the rural areas of low-performing states, opt only for the JSY financial scheme, there is an insignificant effect on the likelihood of children being stunted or underweight. When these mothers decide to opt for both the JSY assistance scheme along the postnatal care of the baby, we find significant effects on the children being less stunted and underweight. Thus only getting financial assistance from the government is not sufficient, the money needs to be spent on the postnatal care of the baby immediately after their delivery to reduce malnutrition among children under five years of age.

​