Population Data Download - Second Modeling Round

MA Population Scenarios (05.06.2003, 320KB Excel)
Techno Garden; Economic Optimism; Local Learning and Fortress World (420KB, Excel)

• all data downscaled to country level
• population data are provided as Excel tables
• preliminary data till final evaluation

Population Scenarios for the MA Storylines

May 23, 2003
Prepared by Brian O’Neill (IIASA), Sergei Scherbov (Austrian Academy of Science; IIASA), Wolfgang Lutz (IIASA; Austrian Academy of Science ) and Anne Goujon (Austrian Academy of Science; IIASA)

Four population scenarios were prepared for the Millennium Assessment Scenarios Working Group, for use in the quantification of storylines. All scenarios were derived from the IIASA 2001 probabilistic projections for the world (see Lutz et al., 2001, Nature), taking into account judgments by the MA Scenarios Group (as expressed at the San Jose meeting in March 2003), as well as by demographers at IIASA, regarding the constraints placed on the population scenarios by the MA storylines.

Table 1 lists the qualitative assumptions about fertility, mortality and migration for each storyline. These assumptions are expressed in terms of categories High/Medium/Low (H/M/L), where these terms are defined not in absolute but in relative terms; specifically, relative to the unconditional expectation for each variable within the IIASA projections. That is, a high fertility assumption for a given region means that fertility is assumed to be high relative to the median of the probability distribution for future fertility in the IIASA scenarios.

Table 1: Fertility, mortality, and migration assumptions by scenario

1. I = Industrialized countries; D = Developing countries; HF = High Fertility countries (TFR>2.1 in year 2000); LF = Low Fertility countries (1.5<tfr<2.1); vlf = very low fertility countries (tfr<1.5).
2. In the IIASA projections, migration is assumed to be zero beyond 2070, so all scenarios have zero migration in the long run.

These assumptions differ in 2 main ways from the assumptions agreed upon in San Jose:

1. Fertility in currently low fertility countries: The knowledge of determinants of long-term trends in fertility in low fertility countries is extremely limited. There is little basis for preferring one set of assumptions over another for a given storyline. While one can construct plausible stories to support the assumptions decided on in San Jose, one could equally well construct plausible stories supporting opposite assumptions (i.e., high instead of low fertility, low instead of high). Given this situation it seemed desirable to devise some overarching rationale for specifying trends for given storylines. It was decided that one useful idea would be convergence. Since Fortress (FT) describes a regionalized, divergent world, and Economic Optimism (EO) a globalizing, convergent world, we decided to apply these characteristics to future fertility. Thus the low fertility countries were divided into two groups (one with Very Low Fertility, one with Low Fertility, see note to Table 1), and fertility assumptions made such that fertility in these two groups would tend to converge in the EO scenario and diverge in the FT scenario. The socio-economic rationale is that in general, the economic conditions and the gender issues (e.g., difficulties for women combining careers and childbearing) would tend to be more similar in the EO world, and regionally differentiated in the FT world. In the Learning scenario, fertility initially follows the FT assumptions, then diverges toward the EO assumptions. In the Technogarden scenario, medium fertility is assumed. Note also that countries are not divided along the traditional lines of Industrialized/Developing countries, but according to current fertility. The one (major) difference this makes is that the region of China and Centrally Planned Asia in the IIASA projections falls in the Low Fertility group.
2. Mortality: assumptions for Technogarden (TG) and Fortress (FT) were adjusted toward higher mortality than originally specified. The TG storyline was originally assigned the same (low) mortality as in Economic Optimism, which seemed inconsistent with the slower rates of economic growth in this scenario, especially for developing countries. Additionally, since the technological emphasis in TG is on environmental technologies, and not necessarily medical technology, there seemed no specific reason that mortality should be as low in TG as it is in EO for industrialized countries either. We therefore assigned medium mortality to the TG scenario. The FT scenario was originally assigned medium mortality, so that it would be higher than TG. We therefore shifted mortality in this scenario to high, to maintain its difference with TG.

To derive quantitative projections consistent with the qualitative criteria, the method employed at IIASA was based on conditional probabilistic projections. Briefly, the steps were:

1. Defining a set of conditional probabilistic projections of fertility, life expectancy, and migration, for each of 13 world regions, which met the above criteria for the FT, EO, and TG scenarios, implemented as follows:
   1. High, medium, and low categories were defined such that they contained those simulations, drawn from the original set of 1000 simulations underlying the IIASA probabilistic projections, that fell into three evenly divided quantiles of the unconditional distribution for each variable;
   2. The basic metric for each variable was a 50-year average (2000-2050); however this was supplemented by additional 50-year averages (2025-2075, 2050-2100) in order to extend the conditional distributions to 2100, and by averages over shorter periods (2000-2015, 2085-2100) to control for boundary effects.
2. Deriving a single deterministic scenario for fertility, mortality, and migration in each of the 13 regions and each storyline, defined as the medians of the conditional distributions for these variables. For the LN scenario, which calls for a world that begins to develop in the same way as FT, and then deviates from it, this procedure was used only for migration. Fertility and life expectancy assumptions were assumed to be identical to the assumptions in the FT scenario up to 2010, and then to deviate smoothly from that scenario to reach a medium path in 2050 and beyond.
3. Projecting future population size and structure based on these four deterministic scenarios for fertility, mortality and migration at the level of 13 world regions;
4. Downscaling the population size results from the 13 IIASA regions to the country level using the same methodology applied to the population assumptions in the IPCC SRES scenarios, as documented on the IPCC TGCIA website (http://sres.ciesin.columbia.edu/tgcia/).

Figure 1 shows results for global population size.

Figure 1: Development of global population according to the four MA scenarios.

The global results are consistent in most ways with the expectations at the San Jose meeting. There are three main exceptions:

1. The range between the lowest (EO) and the highest (FT) scenario is somewhat smaller than may have been anticipated. The range is 8.1 – 9.6 billion in 2050, and 6.8 – 10.5 billion in 2100. These ranges cover 50-60% of the full uncertainty distribution for population size in the IIASA projections. The primary reason that these scenarios do not fall closer to the extremes of the full distribution is that they correlate fertility and mortality: the FT scenario generally assumes high fertility and high mortality, and the EO scenario generally assumes low fertility and low mortality. Both of these pairs of assumptions lead to more moderate population size outcomes.
2. The LN scenario is nearly identical to the FT scenario at the global level over most of the century, even though it is designed to follow the FT scenario only for 10 years and then diverge from it. The reason for this outcome is that (a) the effects of deviations in fertility in the LN scenario do not become apparent in population size for many decades due to population momentum, and (b) both fertility and mortality trends diverge. Thus, although fertility declines in LN relative to FT after 2010, tending (eventually) toward a smaller population size, mortality declines relative to FT as well, tending toward a larger population size. The net result is little difference, especially in the short to medium term.
3. The range of outcomes for one region, North America, is particularly small over all four scenarios, despite widely differing sets of assumptions about input variables. The reason is that when fertility is low in this region, assumptions for mortality and migration tend to offset its effect on population size: mortality is low as well, and migration (which has a substantial influence on population growth in this region) is high. A similar situation holds, in reverse, when fertility is high. Thus the range of population size outcomes is only 426 – 439 million in 2050, and 420 – 540 million in 2100.

The full set of projection results are provided in the form of Excel spreadsheets containing total population size in 5-year time steps over the period 2000-2100 for all countries, and aggregated into 13 IIASA regions, 6 MA regions, and the world. This format is identical to the one used to present the population assumptions used in the IPCC Special Report on Emissions Scenarios, available on the IPCC TGCIA website.As noted in the results file, there are a few minor exceptions: (1) the unit is persons (MA) instead of thousand persons (SRES); (2) Macao is in China and centrally planned Asia (MA) instead of being in Pacific Asia; (3) Three countries are added: Channel Islands in Western Europe, and French Guiana and Saint Lucia in Latin America; (4) There are 5 world regions (MA) instead of 4 (SRES). Further documentation of the methodology, input assumptions, and projection results is in preparation. Questions should be directed to Brian O’Neill (oneill@iiasa.ac.at).

Countries with discontinuities in population size in 2050 (added 05.06.2003)

Countries with discontinuities in population size in 2050, an artifact of the downscaling methodology. Population scenarios for these countries should be used with caution, especially in any context in which the specific population path followed over time is important.