Arguments & Assumptions

Probabilistic Population Projections by 13 World Regions

One can download the assumptions of the uncertainty range of the fertility, mortality, and migration rates used in the IIASA probabilistic population forecasts 2001 (2000 simulations, correlation=0.7 & 0.9) in MS Excel format (IIASA_projections2001.xls).

A more detailed description of the assumptions and the arguments, as well more references, can be found in the book


Future Fertility

Future fertility in today's high fertility countries

The fertility estimates of the world regions for 2000 are highest in SSA with around 5.3, MEA with 4.0, NAF with 3.3, SAS and CAS with 3.1, LAM with 2.6, and PAS with 2.5. All of these fertility levels are significantly lower than in 1995, the base year of IIASA's 1996 projections. The primary basis for the assumptions, that fertility in these regions will continue to fall, is confirmed by the theory of demographic transition, which suggests that once countries enter the secular fertility decline, fertility will continue to fall until levels at or below replacement level are reached. Key elements of the assumptions lie in the timing of the onset of the fertility transitions, the pace of fertility decline, and the level of fertility after completion of the transition. Most of the uncertainty is due to the speed of the fertility decline and to the assumed level of post-transitional fertility. The trends in the means of the regional levels have been defined for periods 2025-2029 and 2080-2084 with interpolation in between. The fertility levels of 2025-2029 are similar to the ones chosen by the UN (2000 Revision), but for 2080-2084 they are assumed to range between 1.5 and 2.0, with lower levels for regions with higher population densities (for discussion see Lutz and Ren 2002). The variances of total fertility rates are assumed to depend on the level of fertility. If the fertility level is above 3.0 we assume that there is an 80 percent chance that fertility would be within one child of the mean; when it is below 2.0, the same probability is attached within one-half child of the mean.

Future fertility in today's low fertility countries

For countries with fertility below subreplacement level, no consistent theory is available. In one sense, forecasting is easier for today's high fertility regions due to the irreversibility of the demographic transition. The possible range of future fertility levels is assumed to be low. In another sense, the task is more difficult in low fertility level countries because we do not know whether fertility will increase or decrease over the coming years. Arguments in favor of assuming further fertility decline and others supporting the assumption of moderate fertility increases are still valid (Lutz 1996, pp. 253-277). During the last few years there has been a debate around the so-called "tempo effect" of fertility, which refers to periods in which the mean age of childbearing is increasing. The period of fertility rates are lower than they would be without this effect (Bongaarts 2002; Kohler and Philipov 2001; Freijka and Calot 2001). The fertility level would be around a mean of 1.7 with an assumed 90 percent interval of 1.2-2.2 for the EEU, WEU, PAO, and the FSU in 2030-2035. Similar distribution around a mean of 1.9 has been specified for NAM.

Future fertility in China

Over the past decades China experienced a precipitous fertility decline. In the 1996 projections, the assumed future fertility level was around a mean of 2.25 with a symmetric distribution of a range of 1.5 and 3.0. Today, there is evidence that fertility may lie below 1.85 (Zhai 2000), and fertility intentions collected in surveys also seem to be falling (CPIRC 2001). Furthermore, the strong educational and urban/rural fertility differentials in China, combined with a strong expected urbanization and improvements of the educational structure (see in Chapters 5 and 8 in Lutz, Sanderson, and Scherbov 2004), allow us to assume rather low future fertility. The assumed future fertility level in China has been defined on an average level of 1.8 and an 80 percent range of between 1.3 and 2.3. These new assumptions are almost half a child lower on average than previously assumed.

Future Mortality

Future mortality in today's higher mortality countries

The developing world saw phenomenal decline in child and adult mortality, mostly due to the reduction of infectious diseases. But in the late 1980s we saw a reversal of this trend for many African countries (see section on AIDS). Possible new unprecedented uncertainties in certain regions arise, in respect to possible impacts of new infectious diseases, worsening environmental situations and fostering old infectious diseases, or significant positive impacts of new medical interventions that could quickly improve life expectancies at a low cost for large segments of the population. Therefore, as a general rule we assume that life expectancy increases by 2 years per decade and that the 80 percent uncertainty interval lies between no gains on the lower side and a 4-year increase per decade on the upper side.

Future AIDS mortality in Africa

The HIV epidemic has spread more rapidly than had been assumed some years ago by the major agencies producing global population forecasts. For instance, the life expectancy assumptions for Botswana were cut by almost half for 2010-2015 by the UN, particularly from 60 years in the 1996 assessment to 32 years in the 2002 assessment. For South Africa, the corresponding assumptions were reduced by as much as 29 years of life expectancy. Though the IIASA 1996 assumptions were more pessimistic than the UN, they turned out to be too optimistic. The forecasts presented here assume a median life expectancy in 2010-2015 of 44 years, and after that initially a slow and later more rapid recovery to 53 years in 2030-2035. The uncertainty range is even wider than those for the other world regions. Over the coming decade the median life expectancy is assumed to decline by 3 years, the 80 percent uncertainty range from a decline of 6 years to constancy. An extensive discussion of modeling the future AIDS mortality dynamics is given in Chapter 7 of Lutz, Sanderson, and Scherbov, 2004.

Future mortality in today's low mortality countries

The uncertainty is not smaller for populations that already have life expectancies above 70 or 80 years. The question of the existence of a fixed biological limit to the human life span is controversial at the moment. The more traditional view is that aging is seen as an intrinsic life process to all cells of the body. Based on this view, a limited life table can be calculated with an average age of death of around 90 and with a maximum life span of 115 years (Duchêne and Wunsch 1991). Olshansky et al. (1990) also assume that an increase beyond 85 years is unlikely because it would require a decline of 55 percent in mortality rates from all causes at all ages. On the other hand, Oeppen and Vaupel (2002) recently showed that the world's highest national life expectancies have increased almost linearly from year to year and shown no sign of leveling off. Because of this great uncertainty, the 80 percent intervals assumed show on the lower boundary a complete stagnation, and on the upper boundary a significant increase of 4 years per decade, and the median path is with a 2-year improvement per decade.

Future International Migration

Immigration trends in Western Europe, USA, Canada, and Australia show remarkable fluctuations. Labor migration within Asia has been remarkable over the last decades. During the 1970s and 1980s a large number of workers went to the oil producing countries in the Middle East. During the late 1980s those migratory streams within Asia have shown significant declines. Due to the volatility of those trends and the greater role that short-term political changes play in both the receiving and sending countries, it is more difficult to hypothesize future migratory streams. For the population forecasts presented here, annual net migration is assumed to be -475,0000 in North Africa, -510,000 in sub-Saharan Africa, -650,000 in Central and -260,000 in South America, -15,000 in Western and -835,000 in Southern Asia, -470,000 in China, and -1,135,000 in Southeast Asia. The distribution of the assumed sending regions to the receiving regions are based mostly on presently observed migratory streams as given in Zlotnik (1996). Model migration schedules by Rogers and Castro (1981) were used to determine the age patterns of migration. For all regions the low value for net migration chosen was zero. For the high end of the uncertainty range, annual migration gains of 2 million in NAM, 1 million in WEU, and 350,000 in PAO, assuming a symmetric distribution with the median lying at half of these levels. Discussion on the increase of international migration movements due to better communication between regions and countries and the associated notion of "environmental refugees" are discussed in length in Chapter 5 (Lutz, Sanderson, Scherbov ).

Print this page

Last edited: 06 November 2012


Sergei Scherbov

Deputy Program Director

World Population

T +43(0) 2236 807 584

International Institute for Applied Systems Analysis (IIASA)
Schlossplatz 1, A-2361 Laxenburg, Austria
Phone: (+43 2236) 807 0 Fax:(+43 2236) 71 313