This paper develops and validates new multivariate methodology to assess the effect of child mortality on period and cohort fertility. Methodology is based on discrete-time survival models that are used to construct multivariate life tables of parity progression. Parity transitions are transitions from the woman’s own birth to her first birth (0-1), from first to second birth (1-2), from second to third birth (2-3), and so on. The basic dimensions of these life tables are age, parity, duration in parity, and child mortality. By multivariate is meant that a life table can be constructed by values or categories of one socioeconomic predictor variable while holding the other socioeconomic predictors constant (socioeconomic predictors may include residence, education, wealth index, religion, etc). These multivariate life tables yield various measures of the quantum and tempo of fertility and child mortality. The methodology also yields a replacement rate, which measures the extent to which parents replace lost children.
The validity of the methodology is tested by applying it to India’s first National Family Health Survey, conducted in 1992/93 (NFHS-1). The application is to both cohort and period data. The cohort analysis pertains to all women age 45-49 at time of survey. The period analysis pertains to the 5-year period preceding the survey, spanning years 1988 to 1992. Tests indicate that basic results agree closely with results from the birth history method and other methods, insofar as comparisons can be made. The replacement rate in the cohort analysis is 0.32 which means about one-third of lost children by cohort of women age 45-49 were replaced by additional births. The replacement rate in the period analysis is 0.50.