Uncertainty in Population Projection for Statistically Less Developed Countries: A Discussion of Estimation Issues Using Data for Four South Asian Countries

Thomas M. McDevitt, U.S. Census Bureau

The literature on uncertainty in population projections has tended to focus on using the historical time series for a parameter to estimate the stochastic component of variation in the parameter on the one hand and, on the other, on using the range of expert opinion about the likelihood of future levels in the components of change to calculate confidence intervals. The distinction between, and relative importance of, the stochastic and bias elements of uncertainty has received much less attention. This paper/poster focuses on the uncertainty in projected fertility for four South Asian countries – Pakistan, India, Bangladesh, and Burma – distinguishing variability attributable to year-to-year variation in estimates (variance) from estimation decision-based variation (bias). The poster draws attention to the fact that estimated uncertainty depends on the selection of fertility estimates used and compares estimated uncertainty with the spread between the United Nations Population Division’s “high” and “low” fertility scenarios.