Population Projections 1. Estimates of future population size, age structure a. Requires assumptions about mortality, fertility, migration 2. Shorter-term estimates a. Depend less on assumptions because people are already born b. Relatively fixed characteristics vs. relatively variable characteristics 1Value of short-term projections 1. Depict likely changes in group composition (e.g., elderly, near-elderly, others) a. Identify, quantify potential policy issues 2. Understand process of cohort replacement a. Effects of cohort size and characteristics 3. Depict historical change across cohorts a. Compare different cohorts at similar points in the life course (age-period-cohort) 2Projections based on one census 1. Forward survive population by age (and sex) a. Use life table survival ratios 2. Assume constant mortality (unchanging SR) 3. Assume homogenous mortality a. No cross-group differences in survival ratios b. Group-specific SR may be available 3Example – Racial composition of US pop age 65+ 1. Inputs a. Census tabulation of race by age (and sex) b. Life table for corresponding year 2. Outputs a. Projected population 10*x years into future (by age) b. Projected population 10*x years into future (by age and race) 4Data – From 2000 Age 2000 Pop 2000 White pop 2000 % White Life table survival ratios Under 5 19175798 12859892 0.67 5 to 9 20549505 13944882 0.68 10 to 14 20528072 14322638 0.70 15 to 19 20219890 14167148 0.70 0.9979 20 to 24 18964001 13064891 0.69 0.9970 25 to 29 19381336 13501773 0.70 0.9936 30 to 34 20510388 14818786 0.72 0.9909 35 to 39 22706664 17031493 0.75 0.9899 40 to 44 22441863 17265995 0.77 0.9880 45 to 49 20092404 15810626 0.79 0.9835 50 to 54 17585548 14213875 0.81 0.9758 55 to 59 13469237 11107247 0.82 0.9642 60 to 64 10805447 8945842 0.83 0.9471 65 to 69 9533545 8040225 0.84 0.9198 70 to 74 8857441 7648193 0.86 0.8784 75 to 79 7415813 6530019 0.88 0.8195 80 to 84 4945367 4408597 0.89 0.7367 85 and over 4239587 3778504 0.89 0.6202 Total 281421906 211460626 0.75 0.3046 5Steps - 1 1. Forward survive 2000 population by age La. x510x5200010-x510x5200010-x5*2010x5L*PS*PP+== b. x510x5*201010-x510x5*201010-x5*2020x5LL*PS*PP+== c. Remember to use appropriate procedure for open ended interval 6Steps – 2 2. Multiply projected population by corresponding racial composition measures Wa. 200010-x5200010-x5*2010x5*2010x5P*PW = b. Calculate age specific population white at x*10 years into future 3. Sum across age ranges of interest, divide by total pop for corresponding age group 7Results ∑∑+=+=∞÷=8565tx58565tx5t65PW%Wxx Year 2010 2020 2030 2040% 65+ White 0.84 0.81 0.78 0.74 8Two censuses 1. Advantage a. Don’t need life table – can use CSRs 2. Disadvantage a. Confounding of mortality and census coverage differentials (and migration) 3. 0xxnxxPPCSR+= 9Steps - 1 1. Construct CSRs 2. Forward survive population by age 3. Multiply projected population by corresponding composition measures 4. Sum across age ranges of interest, divide by total pop for corresponding age group 10Example – educational attainment in Taiwan Data Projected Population in Future Age 1980 1990 CSR 2000 2010 202020 to 24 998,246 979,09825 to 29 905,536 1,012,21430 to 34 608,247 949,834 0.9515 931,615 35 to 39 463,298 839,726 0.9273 938,651 40 to 44 448,575 554,085 0.9110 865,255 848,65845 to 49 434,248 458,082 0.9887 830,272 928,08350 to 54 481,707 422,277 0.9414 521,601 814,529 798,90555 to 59 384,145 410,021 0.9442 432,525 783,951 876,30560 to 64 267,452 419,254 0.8704 367,529 453,976 708,92665 to 69 202,218 302,411 0.7872 322,781 340,497 617,15170 to 74 88,518 182,613 0.6828 286,262 250,944 309,969 75 to 79 45,862 110,233 0.5451 164,850 175,954 185,612 80 and over 25,676 62,650 0.3914 139,150 231,044 257,535 11Use 1980 age-specific proportions w/senior high school education to weight projected populations Projected pop with senior HS or more Age 2000 2010 202020 to 24 25 to 29 30 to 34 35 to 39 40 to 44 442145 45 to 49 375283 50 to 54 194036 416224 55 to 59 118079 354346 60 to 64 71668 168879 36226165 to 69 55841 92956 27895270 to 74 60115 48934 11530875 to 79 33465 30440 5067280 + 17688 41596 47046 12Sum across age range of interest, divide by projected total population 1. e.g., 60+ as in H&C 2. ∑∑+=+=∞÷=8060tx58060tx5t60PHS%HSxxAge/Year 1990 2000 2010 202060+ 0.1858 0.1865 0.2636 0.4109*Compare with figures in the paper *Assumptions about open-ended category 13Transition Rates 1. Useful for analysis of variable characteristics 2. Forward survive population with characteristic x 3. Compare projected number with observed number 4. Difference is net transition 5. Divide difference by average of two populations 6. Result is net transition rate 1415 000,1*]}2/)PP[(D{TR10-tz,10-x5tz,x5tz,x5zx5+÷= 10-tz,10x5tz,x510- tz,10-x510x510- tz,10-x5* tz,x5PP*PS*PP−== tz,x5* tz,x5 tz,x5PPD −=