Beyond Six Billion Forecasting the World's Population (2000) / Chapter Skim
Currently Skimming:
Appendix B: Accuracy of Population Projections from the 1970s to the 1990s
Appendix B: Accuracy of Population Projections from the 1970s to the 1990s
Pages 254-302
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 254... ...
DATA AND MEASURES Data The projections to be covered are from the U.N. Population Division and the World Bank, plus one additional set from the U.S.
|
From page 255... ...
The 1972 World Bank forecast does not provide life expectancy or infant mortality, and in addition reports results for decades rather than 5-year periods.
|
From page 256... ...
will therefore be A peit = (Pit - Pit ) / Pit where P stands for actual population and P for projected population.
|
From page 257... ...
In this case, however, raw differences rather than proportions are taken for total fertility, life expectancy, and infant mortality, given that component rates are already expressed in relation to population. Migration assumptions are similarly assessed but require some explanation.
|
From page 258... ...
Initial life expectancy and initial net migration are also used in some analyses. · Developing country and industrial country (or developed country)
|
From page 259... ...
(It is reexamined below as part of the multivariate analysis.) Although the absolute errors in these projections may appear large, the projections are a considerable improvement over holding population or population growth rates constant.
|
From page 262... ...
Using these data as criteria for a different agency's projections, rather than the other agency's own current estimates, could produce apparent error that is actually disagreement between agencies about historical demographic trends. For instance, the mean base absolute proportional error in the 1983 World Bank forecast, in Table B-1, is 0.034.
|
From page 263... ...
Some Sources of Error The variability in country projections is demonstrated even more clearly in Figure B-2, which gives percentiles of the distribution across countries. Positive bias in the base population is confirmed as more serious than negative bias.
|
From page 264... ...
Demographic upheavals of this sort, or demographic quakes, might be identified from sharp increases or decreases in the annual population growth rate, of at least 2.5 percentage points from one 5-year period to the next. (This quantity is equal to twice the standard deviation for such period-to-period growth-rate changes.)
|
From page 265... ...
Since quake-hit countries are relatively few, eliminating them affects overall average absolute proportional error only slightly though consistently, reducing it by 1-2 percentage points across projection lengths. Regional variation produces smaller but also important variation in
|
From page 266... ...
FIGURE B-4 Proportional error and absolute proportional error by the occurrence of demographic quakes and projection length. projection error.
|
From page 267... ...
The adjustment reduces proportional error and absolute proportional error, but much more for shorter projections than for longer ones. For 5year projections, the reduction in mean absolute error is 50 percent or more, and this is fairly consistent across forecasts.
|
From page 268... ...
Part of the reason for this is that the growth rates evaluated are cumulative, with growth rates for longer periods incorporating the rates for shorter periods within them. However, the current data still show a consistent increase in error, even within each forecast, with projection length.
|
From page 269... ...
With only eight cases, a general statement cannot be made about the comparative accuracy of this alternative methodology. ACCURACY OF PROJECTED COMPONENT RATES Component rates can be examined in the same manner as population, comparing projected rates from a specific forecast for periods up to 19952000 with rates reported by the U.N.
|
From page 270... ...
The equivalent absolute error in the crude birth rate across all projection lengths, calculated from reported crude birth rates, is 3.4 per thousand. This amount is the average distortion in the population growth rate, in one direction or the other, produced by fertility error.
|
From page 271... ...
Absolute error in projected life expectancy, in contrast, shows a clear increase with projection length, growing from 2.3 years of life expectancy in 5-year projections to 4.0 years of life expectancy in 20year projections. The rate of increase in absolute error with projection ~ J
|
From page 272... ...
Absolute error in life expectancy appears to show some decline in the most recent forecasts. On average across projection lengths, the equivalent absolute error in the crude death rate is 1.9 per thousand, meaning that mortality error has on average half the effect on population growth of fertility error.
|
From page 273... ...
Absolute error in the net migration rate ranges from 3.7 to 5.1 per thousand, except for the last period, with no obvious trend. Across all projection lengths, the average is 3.9 per thousand, which is slightly greater error than there is in the crude birth rate and double the error in the crude death rate.
|
From page 274... ...
The error for developing countries is much more difficult to characterize, but some overall underprojection of net emigration is possible. 3 2 o -2 3 Europe/FSU US/Australia/Japan Latin America/Caribbean Asia Mideast/North Africa Sub-Saharan Africa .~ ~ \ / \~W \ · ~ \ " \ \ ' ' \ \ / i' X i, / ' \ \ ~ \ \\ / \ (% \ , ~ \ \ , I` / \ , ~ i, i, ~ , \ I ~ , I \l I I I / \ , -- 'N; 1975-1980 1980-1985 1985-1990 1990-1995 1995-2000 FIGURE B-9 Error in projected net migration rate, by region and target period.
|
From page 275... ...
Absolute error in projected component rates increases with projection length for total fertility and life expectancy, but not that strongly for infant mortality, and not at all where migration is concerned. Base error appears to be important for all the component rates.
|
From page 276... ...
276 APPENDIX B TABLE B-6 Analyses of covariance for absolute proportional error in projected population Source of variation SS df MS F p< Covariates 33.463 2 16.732 1103.17 0.000 Base population error 0.279 1 0.279 18.41 0.000 Base error2 3.130 1 3.130 206.39 0.000 Main effects 24.802 19 1.305 86.07 0.000 Projection length 10.940 5 2.188 144.26 0.000 Year projected 0.143 5 0.029 1.89 0.093 Recent census 0.022 1 0.022 1.46 0.227 Demographic quake 3.429 1 3.429 226.10 0.000 Initial population 0.962 1 0.962 63.46 0.000 Initial fertility 0.021 1 0.021 1.41 0.236 Region 1.878 5 0.376 24.77 0.000 Explained 58.265 21 2.775 182.94 0.000 Residual 74.317 4,900 0.015 Total 132.583 4,921 0.027 Covariates 33.463 2 16.732 1109.97 0.000 Base population error 0.279 1 0.279 18.53 0.000 Base error2 3.130 1 3.130 207.66 0.000 Main effects 25.303 22 1.150 76.30 0.000 Projection length 11.500 5 2.300 152.58 0.000 Forecast 0.644 8 0.080 5.34 0.000 Recent census 0.017 1 0.017 1.16 0.282 Demographic quake 3.437 1 3.437 228.00 0.000 Initial population 1.041 1 1.041 69.09 0.000 Initial fertility 0.028 1 0.028 1.86 0.172 Region 1.830 5 0.366 24.28 0.000 Explained 58.766 24 2.449 162.44 0.000 Residual 73.817 4,897 0.015 Total 132.583 4,921 0.027 Covariates 33.463 2 16.732 1104.15 0.000 Base population error 0.279 1 0.279 18.43 0.000 Base error2 3.130 1 3.130 206.57 0.000 Main effects 24.913 22 1.132 74.73 0.000 Year projected 11.110 5 2.222 146.64 0.000 Forecast 11.051 8 1.381 91.16 0.000 Recent census 0.017 1 0.017 1.15 0.284 Demographic quake 3.437 1 3.437 226.81 0.000 Initial population 1.041 1 1.041 68.73 0.000 Initial fertility 0.028 1 0.028 1.85 0.173 Region 1.830 5 0.366 24.15 0.000 Explained 58.376 24 2.432 160.52 0.000 Residual 74.206 4,897 0.015 Total 132.583 4,921 0.027 Note: The dependent variable and base error are both cube roots of absolute error.
|
From page 277... ...
By region, this multivariate exercise confirms the patterns shown earlier: absolute error is lower in industrial countries and substantially higher for the Middle East and North Africa. Projection error can therefore be expected to be greater when base estimates are poorer; for projections of greater length; possibly in earlier rather than in the latest forecasts; when demographic quakes upset the population growth trend; for smaller populations; for developing coun
|
From page 278... ...
-3.46 -1.09 -3.67 360 tries; and particularly, in the last few decades, for the Middle East and North Africa. Absolute proportional error in base population estimates was analyzed separately to see if similar factors accounted for it.
|
From page 279... ...
By region, the pattern is similar to that for projected population, except that population estimates are not quite as bad for the Middle East and North Africa and clearly worse for Sub-Saharan Africa. What appears to characterize error in base population estimates, in contrast to error in projected population, is a greater influence of the specific period estimated, the mid-1980s apparently having been a difficult period during which to make current population estimates.
|
From page 280... ...
Accounting for Error To account for error, regressions were run for proportional bias in projected population (not for absolute error) , using proportional bias in initial population and bias in total fertility, life expectancy, and net migra
|
From page 281... ...
Using these regressions, the variance in projection error can be partitioned into that explained uniquely by each factor, as well as that jointly explained. This requires running several regressions leaving out factors one at a time, to determine the marginal contribution that each factor TABLE B-9 Regressions to account for proportional error in projected population, by projection length Variable B (t)
|
From page 282... ...
5 10 I Joint | effects | Migration | (Arnold) | Mortality | Fertility | Base | population FIGURE B-10 Percent of variance in proportional error in projected population explained by proportional error in base population and error in component rates, by projection length, using alternative migration estimates.
|
From page 283... ...
Base error is by far the most important factor producing error in projected total fertility and life expectancy, accounting in both cases for a sixth of the total variance, or half of all the explained variance. It is also a significant factor in projected net migration rates, but in this case, the
|
From page 284... ...
The effects of projection length appear to reinforce this point. Projection length is responsible for substantial error in total fertility, some error also in life expectancy, but none in net migration.
|
From page 285... ...
Table Bell summarizes the analyses of variance, and Table B-12 gives adjusted deviations. Base error appears to be higher if total fertility is initially higher, if life expectancy is initially lower, and if the net migration rate is initially more positive just as is the case with error in projected component rates.
|
From page 288... ...
The relatively large error in projected net migration rates for the Middle East and North Africa is substantially reduced, presumably when demographic quakes are taken
|
From page 289... ...
Africa Sub-Saharan Africa 289 Projected life expectancy Basellife expectancy Unadjusted Adjusted -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 FIGURE B-12 Deviations from mean absolute error in projected and base life expectancy, by region (unadjusted and adjusted for other effects)
|
From page 290... ...
Africa Sub-Saharan Africa APPENDIX B Projected net migration rate Base net migration rate Unadjusted Adjusted 2 3 4 FIGURE B-13 Deviations from mean absolute error in projected and base net migration rate, by region (unadjusted and adjusted for other effects)
|
From page 291... ...
These estimates are mostly between 5 and 20 percent of equivalent mean country absolute errors (Table B-14~. World projections therefore appear relatively precise.
|
From page 293... ...
It does however increase fairly clearly from 1990 to 1995 to 2000 at every projection length, indicating that world population in the l990s became increasingly difficult to project. Explaining Increased Error The increase in world absolute error in the l990s is not explained by an increase in average country error.
|
From page 294... ...
15 20 25 FIGURE B-15 Proportion of country error unoffset in world projections across six exercises, by projection length and target year.
|
From page 295... ...
·~. 1980 1985 1990 1995 2000 295 FIGURE B-16 World total fertility rate in various forecasts, by target period.
|
From page 296... ...
Mortality appears to have reached a similar turning point around 1990, with progress in life expectancy moderating somewhat around that date (Figure B-17~. However, the effect on projection accuracy is ambiguous, since some forecasts projected higher life expectancy and some projected lower life expectancy.
|
From page 297... ...
First, the base demographic estimates from which the projection starts may be wrong, producing subsequent errors. The base population may be misestimated, as may be the base levels of fertility, mortality, and net migration.
|
From page 298... ...
A misspecified component rate can cause increasing error in projected population as a projection lengthens, and in addition projected levels of fertility and mortality become increasingly poorly specified. The third major explanation for projection error is unexpected events that substantially alter demographic prospects.
|
From page 299... ...
Base population error, to begin with, is substantial, but appears to go in both directions, so that country populations are neither systematically over- or underestimated. Mortality has been systematically underprojected for this region and accounts for proportionally more error in projected population than in other regions.
|
From page 300... ...
Base estimates are responsible for much projection error, especially in the short term, when projection results are most useful for planning. Base estimates for population, age structure, and component rates can be improved, in general, through better data from censuses, surveys, and vital registration and through improved methods of estimating demographic parameters.
|
From page 301... ...
Vu, and R.A. Bulatao 1991 Africa Region Population Projections; Asia Region Population Projections; Latin America and the Caribbean Region Population Projections; Europe, Middle East, and North Africa Region Population Projections, 1990-91 Edition.
|
From page 302... ...
Bos, and R.A. Bulatao 1988 Africa Region Population Projections; Asia Region Population Projections; Latin America and the Caribbean Region Population Projections; Europe, Middle East, and North Africa Region Population Projections, 1988-89 Edition.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.