The Own Children Fertility Estimation Procedure: A Reappraisal

Christopher Avery, Harvard Kennedy School
Travis St Clair, Harvard Kennedy School
Michael Levin, Harvard University
Kenneth Hill, Harvard University

We estimate Total Fertility Rates (TFR’s) using the Own-Child and Full Birth History methods for 56 countries using DHS data. Although the Household Questionnaire (HQ) and Individual Questionnaire (IQ) include exactly the same set of women, we find that the Full Birth History method yields significantly higher estimates of TFR than does the Own-Child method. We find two systematic sources of differences in these explanations. First, we find systematic inconsistencies in age coding across the two questionnaires. Second, we find evidence of systematic sampling bias in the IQ file; women who appear in the HQ file but not in the IQ file tend to have relatively fewer children. Each of these differences yields an estimated increase of 2% in TFR for the Full Birth History method relative to the Own-Child TFR. These findings suggest that the Own-Child method is probably at least as accurate as the Full Birth History method.