Advances in population projections

Over the last two decades significant progress has been made in the development of probabilistic population forecasting methods, and many applications to countries and global regions are now in evidence. Unfortunately little interest has been shown in the extension of these methods to subnational areas. Given that forecast error is inversely related to population size, coupled with the fact that much planning occurs at the local and regional scale, the need to quantify forecast uncertainty for subnational regions is arguably even greater than at the national scale. This paper presents a regional probabilistic population and household projection for Sydney, Australia’s largest metropolitan region and home to about 4.6 million people. A bi-regional framework is adopted consisting of the region of interest and the rest of the country; predictive intervals for fertility, mortality, internal migration, international migration and living arrangements are formulated on the basis of time series models, past errors and expert judgement. The results demonstrate the considerable extent of forecast uncertainty for the region, and reveal how the official high-low projection range provides a misleading indication of uncertainty. The paper concludes by discussing how probabilistic forecasts might be applied in metropolitan regional planning.

This paper aims to forecast the probability of progression to the secondary education level for Brazilian men and women aged 25-59. The authors used an age-period-cohort model and a time-series approach to model and forecast future trends in period and cohort effects. To account for uncertainty, this study utilized a scenario approach. The authors carried out separate models for males and females in order to account to sex differentials in the progression probability. Data were selected from the National Household Survey in Brazil (PNAD), which offers nationally representative data from 1981 to 2011. This study concludes that there will be improvements in this indicator for both men and women, and that the rate of improvement will be higher for older adults because they start at a lower probability level. Also, these results show that the probability of progression to the secondary education level is slightly lower for men, even though both men and women will experience improvements in this indicator.

During the past 10-15 years a number of papers have surfaced that propose new methods for projecting populations and fertility. Given the wide array of such projections, it appears that the time is right to step back and take stock of the ‘state of the art’ and to critically evaluate the relative contributions of each method. We will analyse these new methodologies and attempt to provide an objective evaluation of their contributions to improving the utility of population and fertility forecasting. In this vein, we critically evaluate at least nine new publications which have driven forward the frontiers of population forecasting using a set of objective criteria for evaluating fertility and population projections which we will design. The main criterion that we are proposing to judge whether a new fertility or population projection method is an improvement, is whether it utilizes research that has revealed a heretofore unknown or not well understood mechanism influencing a fertility status or trend, or a new mechanism influencing population change. The paper will conclude with an overview of the type of fertility and population projections needed or requested by different types of users and an evaluation of whether the supply meets the demand.

There is increasing concern about the lack of accuracy in population projections at national levels. A common problem has been the systematic under-estimation of improvements in mortality, especially at older ages, resulting in projections that are too low. In this paper, we present a method that is based on projecting survival rather than mortality, which uses the same data but differs technically. In particular, rather than extrapolating trends in mortality, we use trends in life expectancy to establish a robust statistical relation between changes in life expectancy and survival using period life tables. We test the approach on data for England Wales for the population aged 50 and over, and show that it gives more accurate projections than official projections using the same base data. Using the model to project the population aged 50 and over to 2020, our method suggests nearly 0.6 million more people in this age group than official projections.

This paper presents the core methodological ideas, empirical assessments and applications of extended cohort-component approach (known as ProFamy model) to simultaneously project households, living arrangements and population age structure/sizes at sub-national level. Comparisons of projections from 1990 to 2000 using this approach with census counts in 2000 for each of the 50 states and DC show that 68.0, 17.0, 11.2, and 3.8 percent of absolute percent errors are <3.0%, 3.0-4.99%, 5.0-9.99% and 10.0%, respectively. Another analysis compares average forecast errors between extended cohort-component approach and still-widely-used classic headship-rate method, by projecting number-of-bedrooms-specific housing demands from 1990 to 2000 and then comparing with census counts in 2000 for each of the 50 states and DC. The results demonstrate that, as compared to extended cohort-component approach, headship-rate method produces substantially more serious forecast errors. Illustrative projections from 2000 to 2050 for each of the 50 states, DC, six counties of Southern-California, and Minneapolis-St. Paul Metropolitan Area are reported. Among many interesting outcomes of projections with medium, low and high bounds, the aging of American households over the next few decades across all states/areas is particularly striking. Finally, the

Recently, the United Nations Population Division adopted a new method for projecting total fertility (TF) for all countries. The new projection method was well received but raised discussion about the model assumption that in the long run, the TF will oscillate around the approximate replacement level of 2.1 for all countries. In this paper, we investigate a modified TF projection model, whereby the ultimate fertility levels are country-specific and estimated using a Bayesian hierarchical model. Expert opinion is incorporated into the model by setting the upper bound on the ultimate fertility level to 2.1. Under the proposed model, ultimate fertility levels are smaller though within 0.25 child of the current UN projection for most low fertility countries, and 1.9 (80% projection interval 1.6-2.3) for countries that have not yet completed the fertility transition, compared to 2.1 (1.8-2.4) for the existing method.

One major hurdle for the Brazilian economy relates to the availability of qualified workforce in key occupations. This paper contributes in this matter presenting a demographic projection of engineering workforce availability in Brazil up to 2020.
The projection method we have used adapts the cohort-component method to a simplified model of entries and exits in the labor market. Data from five different public databases available in Brazil were used to run the projections on yearly basis. The method is rather flexible and can be used to project almost any population group with higher education degree from different backgrounds, by sex and age. We draw four scenarios that differ in terms of possible growth rates to be observed in the number of student entries, including a constant enrollment number (CER) variant.
The results suggest that Brazilian labor market would have between 1.6 and 2.3 million people holding a degree in engineering fields in the year 2020. Finally, our findings suggest that the claimed lack of engineering workforce in the country might not be a matter of purely quantitative supply, but rather of education quality and geographical concentration of engineering schools and engineering workforce.

In this paper, we develop a Bayesian cohort population projection model that incorporates model uncertainty. We first argue that a Bayesian approach is a more natural framework for incorporating various forms of uncertainty in probabilistic projections. Second, we demonstrate the differences that arise from choosing different Lee-Carter type models for fertility, mortality, immigration and emigration in terms of forecasted age patterns and their associated measures of uncertainty. Third, we incorporate this information into a cohort component projection model and use Bayesian model averaging techniques to produce a model-averaged population forecast for the United Kingdom by age and sex. We end the paper by discussing the merits and flexibility of a Bayesian cohort component projection model and highlight some areas where this work could be extended.

The United Nations Population Division releases country fertility estimates and projections every two years, currently using the model of Alkema et al (2011, Demography) for total fertility rate (TFR). This Bayesian hierarchical model produces a predictive distribution of TFR for each country. We extend this model to allow probabilistic projection of the TFR for any set of countries, such as a region or trading bloc. We model the correlation between country TFRs that is not captured by the original model as a linear function of time invariant covariates, namely whether the countries are contiguous, whether they had a common colonizer after 1945, and whether they are in the same UN region. This correlation structure is incorporated into the original model's error distribution and is shown to improve the calibration of predictive intervals for the future TFR of regions.

In classical population projection it is inherently assumed equality between census survival ratios and the corresponding life table survival ratios, which hold true only when the population under study is stationary or stable. The proposed technique makes use of ten-year cumulative census survival ratios which run more smoothly than the conventional census survival ratios (as shown by Coale and Demeney, 1967). These ten-year cumulative census survival ratios have been estimated from the corresponding ten-year cumulative life table survival ratios (defined by the ratio Tx+10/Tx in life table terminology) through a formula developed by Lahiri (2003) under generalized population model assuming that the age-specific growth curve follows a second-degree polynomial. The technique proposed here has been applied to project the Indian populations in 2001 and 2011 taking 1991 and 2001 census age-returns as bases respectively. The projected age-sex populations in 2001 and 2011 are sufficiently close to those of the 2001 enumerated population and 2011 expected population respectively. The technique can also be extended for long-term adult population projection by estimating the prospective growth rates over time.