Tadandjoa Kolani - Research

Research Focus

My research focuses on structural econometrics, causal inference (point and partial identification), robust inference, and applied microeconometrics, with applications in health, population, and development economics. I am particularly interested in policy evaluation and the theoretical foundations of econometric identification.

Published Articles

Heterogeneous Returns to Antenatal Care: Evidence from Childhood Immunization in Senegal
Economics & Human Biology, 2026.

Abstract

Antenatal care (ANC) serves both as a medical input and as a source of health information in the child health production function. We use seven waves of the Senegal Demographic and Health Surveys, matched to regional health-facility data, to estimate the effect of ANC adherence on childhood vaccination outcomes. We address endogeneity arising from unobserved heterogeneity in ANC utilization using a two-stage copula-based control-function approach. We find that ANC adherence has no effect on the composite measure of full immunization, but significantly increases uptake of two injectable vaccines, the three-dose diphtheria-pertussis-tetanus vaccine (DTP3) and measles, by 8.2 % and 6.3 % points, respectively. These effects are concentrated among mothers with no schooling or only primary education. We also find no robust evidence that the marginal returns to ANC adherence differ by child sex, fertility intention, or ethnicity.

Working Papers

Help Me! If You Help Him, Are You Really Helping Him? (with Liliane Bonnal and Pascal Favard)
Draft available soon.
Estimating Informal Care Effects beyond Instrumental Variables: Evidence from Long-Term Care in Europe (with Liliane Bonnal and Pascal Favard)
Submitted.

Abstract

Informal care (IC) is central to long-term care (LTC) systems, yet estimating its effects remains difficult because care arrangements are endogenous. This paper examines whether a copula-based control function approach can estimate IC effects without strong instruments while reducing the risk of overstated effects from instrumental-variable (IV) strategies. We study IC and LTC outcomes among older adults living alone in Europe, where commonly used child-based instruments are intuitive but not consistently relevant across care types, gender, or institutional contexts. Using SHARE Wave 9 data on non-institutionalized individuals aged 65 and over in 25 European countries, we implement a two-stage copula-based control function estimator (2sCOPE), which does not require excluded instruments, and benchmark it against two-stage residual inclusion IV estimates. Informal domestic care reduces formal care use by 19.5 percentage points, while informal personal care reduces formal personal care use by 5.8 percentage points. By contrast, IV estimates imply an effect on formal domestic care nearly twice as large, while instruments for informal personal care are too weak to support reliable IV estimation. IC also lowers depressive symptoms and loneliness, with heterogeneity by gender and European region. Overall, 2sCOPE helps refine estimates of IC effects when IV strategies are limited. Our results suggest that, under fiscal pressure from population ageing, policies promoting ageing in place should target the groups and contexts in which IC reduces FC use or improves mental well-being, thereby enhancing LTC system efficiency.
Family Size, Compliance Costs, and Childhood Vaccination (with Cornel OROS)

Abstract

Childhood vaccination is a high-return early-life human capital investment whose monetary price is close to zero in many developing countries, yet many households fail to complete the schedule. We argue that family size raises the shadow cost of completion and develop a framework in which additional children make schedule compliance more difficult. Using household data from Senegal and twinning among older siblings as a source of plausibly exogenous variation in family size, we estimate that one additional surviving older sibling reduces the probability of full vaccination by about 10.5 percentage points. The effect is zero for vaccines administered at birth, but negative for vaccines requiring repeated visits or later returns, consistent with a compliance-cost mechanism. Effects are larger where completion costs are more likely to bind, including among children whose mothers have no formal education and those living in rural areas. We also find evidence consistent with positive selection into twin births, which attenuates the estimated negative effect of family size. Our findings show that family size can constrain early-life human capital accumulation even when the investment is financially accessible.
When More Children Hurt or Help: Family Size, Household Decision-Making, and Childhood Vaccination (with Liliane Bonnal, Pascal Favard and Cornel OROS)
Active working paper.

Abstract

This paper studies how family size affects childhood vaccination through two competing mechanisms: completion costs and learning by doing. We use household data from Senegal and Nigeria and exploit twinning among older siblings as a source of variation in sibship size. The effect of family size is sharply context-dependent. In Senegal, an additional older sibling reduces vaccination completion, especially among children of non-educated mothers, consistent with a completion-cost mechanism. In Nigeria, by contrast, family size increases vaccination, but only in households where the husband or partner is reported as the sole decision-maker, and especially when the mother has some education. These findings show that larger families generate both constraints and capacities, and that the quantity–quality trade-off in child health is not universal.

The Formal–Informal Care Relationship: Heterogeneous Effects by Care Needs, Social Capital, and Psychosocial Well-Being (with Liliane Bonnal, Pascal Favard, and Cornel Oros)

Work in Progress

Female Autonomy and Child Health Outcomes: Evidence from Developing Countries
Not All Non-Use Is the Same: Contraceptive Intentions and Early-Life Health Investments
Fertility, Intergenerational Transmission, and Child Health
Identification in Recursive Multivariate Probit Models: Do Exclusion Restrictions Really Matter?