 |
 |
|||
 |
|
|||
| |
|
|||
|
|
|||
Technical Notes
2000 Census-based Population Projections Technical Notes 
The 2000 Census of Population and Housing (Census 2000) was conducted by the National Statistics Office from May to July 2000, with May 1, 2000 as the reference date. President Gloria Macapagal-Arroyo signed Proclamation No. 28 on April 18, 2001, declaring the results of the Census 2000 as official for all purposes, and placing the population count of the Philippines as of May 1, 2000 at 76,498,735 persons. This number was later adjusted to include the census counts for some enumeration areas that were inadvertently overlooked during the processing of census forms. The final population count as of May 1, 2000 is 76,504,077 persons. This number was forward-survived to July 1, 2000 to derive the national base population for the 2000 Census-based population projections. Similarly, for each of the regions and provinces, the population as of July 1, 2000 was computed to obtain the base populations that will be used for the regional and provincial projections. This procedure is in line with the conventional practice, when preparing population projections, to use July 1 as the reference date. This is because the midyear population normally represents the average population for the year.
Prior to forward-surviving the May 1, 2000 population counts to July 1, 2000, an evaluation of the age-sex distribution of the provincial populations based on the Census 2000 was carried out. Indexes used in assessing the quality of age-sex data, such as UN Age-Sex Accuracy Index, Myer’s Index, Whipple Method and Bachi Method, which are designed to measure age heaping or digit preference, were calculated. These indexes indicate minimal age heaping and digit preference.
Census survival ratios for 1980-1990, 1990-1995 and 1995-2000, and ratios of population aged 0-4 to population aged 5-9 in the same year were also calculated, and age-sex pyramids were generated, using data for the Philippines and all provinces for census years 1980 to 2000, in order to allow comparison and scrutiny of trends.
In 58 of the 79 provinces, population aged 0-4 is smaller than the population aged 5-9. In order to determine if this is demographically plausible, the fertility rates and migration rates for these provinces were examined. Level of fertility is still generally high so that in the absence of massive migration, the age-sex structure of almost all provincial populations is expected to be one that shows broad population base, with the age group 0-4 comprising the largest percentage.
A possible undercount of children below 1 year of age was concluded because it has been widely observed that infants tend to be missed during censuses by respondents when reporting the members of their households. Using the estimates of total fertility rate (TFR) and age-specific fertility rates (ASFR) for the Philippines from the 1998 and 2003 National Demographic and Health Surveys, the population aged 0 as of May 1, 2000 was estimated. The calculation involves estimating the total number of births during the year prior to May 1, 2000 using the interpolated TFR and ASFR for that year. Applying the sex ratio at birth equal to 1.06, the male and female births were derived. These were forward-survived to age 0 by multiplying these births by the life table survival ratios, for males and females, from birth to age 0. These survival ratios were taken from the 1995 Gender-Specific Life Tables for the Philippines, Its Regions and Provinces by Dr. Cabigon and Fr. Flieger.
The difference between the estimated P 0 and the P 0 enumerated in the Census 2000 is 146,582. This is the estimated undercount in P 0 at the national level during the Census 2000. This number was added to the July 1, 2000 population aged 0 (using the population growth rate from May 1, 1990 to May 1, 2000). This number was distributed to all provinces except for NCR and provinces in ARMM based on their population size.
The age-sex distributions of the provinces in ARMM were smooth for age-misreporting and possible error in census coverage that could vary by age and sex. The AGESMTH program in the Population Analysis Spreadsheet (PAS) software was used to smooth the age-sex distributions of the ARMM provinces. From among the smoothing procedures provided by AGESMTH, the Strong Moving Average Method was selected because this is the only method that smoothes the population under 10 years of age and for ages 70-79. The smoothing was done for each sex separately. After the age-sex distributions of ARMM provinces have been smoothed, the population counts for each age-sex groups of all provinces were added to derive the final baseline population by age and sex for the Philippines.
Derivation
of Fertility Estimates
Fertility is one of the demographic processes that affect population change. This is vital in preparing population projections, specifically, using the cohort-component method with the PEOPLE software. The Inter-Agency Working Group on Population Projections (IAWGPP) established the Small Working Groups (SWGs) on Fertility, Mortality, Migration and Base Population to study the necessary inputs of these components of population projections.
The SWG on Fertility explored all available data to come up with the most robust estimates of the current level of the total fertility rate (TFR) and the pattern of fertility (age specific fertility rates, ASFRs) of Filipino women in the year 2000. The availability of many methodologies for fertility estimation and various data sources made the assessment and estimation more complicated because of the different results. The SWG tried to use the results of the 2000 Census of Population and Housing (CPH). However, the fertility pattern of ASFRs did not reveal the expected configuration that births among women aged 45-49 equals zero. The data showed otherwise.
The P/F ratio method was applied to the 2000 CPH, the 2000 Vital Registration Records and the 2000 Family Planning Survey (FPS) round. Data were examined and graphs were prepared for all provinces. The component provinces of ARMM did not show good results. The Palmore and Rele methods revealed lower estimates. Moreover, estimates of TFR from Palmore 2000 could not be used to evaluate values of the base population because the main input for the Palmore method came from the same source. The TFR equal to 3.50 from the 2003 NDHS was used.
Table 1 below shows different TFRs derived from survey, census and vital registration system by age group.
Table 1. ASFRs and TFRs by Data Source and by Age Group: 2000
AGE GROUP |
2000 |
||
FPS |
Vital Registration |
CPH |
|
15-19 |
0.02 |
0.04 |
0.03 |
20-24 |
0.12 |
0.17 |
0.13 |
25-29 |
0.17 |
0.20 |
0.19 |
30-34 |
0.14 |
0.16 |
0.17 |
35-39 |
0.10 |
0.10 |
0.14 |
40-44 |
0.05 |
0.04 |
0.09 |
45-49 |
0.01 |
0.01 |
0.07 |
TFRs |
3.04 |
3.60 |
4.07 |
The next procedure done was to interpolate the 2000 TFRs at the national and regional levels from the 1998 and 2003 National Demographic and Health Survey (NDHS). For the provincial level, the ratio of the provincial to the regional 1990 Palmore values multiplied by the interpolated regional TFR for 2000 were derived, except for ARMM.
Finally, the IAWGPP adopted en toto the regional estimates of TFRs and ASFRs based from the 2003 NDHS since these estimates refer to the past 0-4 years before the survey date. The table below shows the base TFR and ASFR used.
Table 2. Age-Specific Fertility Rate and Total Fertility Rate, Philippines: 2003 NDHS
Age Group |
Age-Specific Fertility Rate |
15-19 |
0.053 |
20-24 |
0.178 |
25-29 |
0.191 |
30-34 |
0.142 |
35-39 |
0.095 |
40-44 |
0.043 |
45-49 |
0.005 |
Total |
0.707 |
Total Fertility Rate |
3.54 |
Fertility Assumption (National Projections)
Variations in the projected TFRs up to 2050 were dictated by the differences in the timing when replacement level of fertility will be achieved, that is when the Net Reproductive Rate (NRR) equal 1. For the LOW series (rapid pace of fertility decline), NRR = 1 was targeted for the year 2030. For the MEDIUM series (moderate pace of fertility decline), NRR=1 was targeted for the year 2040 and for the HIGH series (slow pace of fertility decline), 2050.
Using the PEOPLE software the fertility estimates were projected to follow an exponential curve for all assumptions. Shown in Table 3 are the projected TFRs for the Philippines under varying assumptions of fertility mentioned above.
Table 3. Projected Total Fertility Rates in the National
Projections
Under Varying Assumptions
Period |
LOW |
MEDIUM |
HIGH |
2000-2005 |
3.37 |
3.41 |
3.44 |
2005-2010 |
3.07 |
3.18 |
3.25 |
2010-2015 |
2.79 |
2.96 |
3.07 |
2015-2020 |
2.54 |
2.76 |
2.90 |
2020-2025 |
2.31 |
2.57 |
2.74 |
2025-2030 |
2.10 |
2.39 |
2.58 |
2030-2035 |
1.91 |
2.23 |
2.44 |
2035-2040 |
1.73 |
2.07 |
2.31 |
2040-2045 |
1.58 |
1.93 |
2.18 |
2045-2050 |
1.43 |
1.80 |
2.06 |
Note: NRR=1 was targeted for the year 2030 for the low series, 2040 for the medium series and 2050 for the high series
The Life Table
The life table is the best analytical tool for the mortality analysis and source of mortality indicators. It comprises of a set of functions representing a synthesis of age-specific mortality observed in the population during a given period. It requires data on deaths (numerator) and population (denominator) by age and sex at this given period to generate such set of functions.
The 2000 constructed life tables for the Philippines at the national, regional and provincial levels are the basic source for the baseline estimates for mortality using the expectation of life at birth (e0) function for the 2000 Census-based population projections.
The 2000 baseline population by age and sex served as the denominator for the national, regional and provincial levels. Unfortunately, the basic problem with the numerator is that death statistics in the Philippines are incomplete, causing a methodological difficulty in constructing life table. Hence, several steps had to be undertaken before the final number of deaths by age and sex was obtained at the required national and sub-national aggregations to calculate the needed life tables. The steps are as follows:
Estimation of the level of completeness of death registration
The first step is to correct the registered deaths, which are in five-year age groups and by sex, for ages five years or older with the application of four techniques of estimating the level of completeness of death registration for ages five years and over. These are Brass’s growth balance method (1975), Courbage and Fargues’s method (1979), Preston and Coale’s method (1980) and Gray’s method (1986). All four procedures require the same type of data, that is, registered deaths and total population by five-year age group and by sex. The different methods have their assumptions, strengths and weaknesses, as summarized below:
Method |
Assumptions |
Strengths/Weaknesses |
Brass |
Assumes that the degree of completeness of death registration is more or less the same at all ages after childhood (or at ages 5 or over) Assumes that the population is stable and closed |
Not very sensitive to age-misreporting but highly sensitive to departure from the closed and stable population assumption and constant under-registration of deaths by age after childhood |
Preston and Coale |
Assumes that the degree of completeness of death registration is more or less the same at all ages after childhood (or at ages 5 or over) Assumes that the population is stable and closed |
Fairly robust to violation of the stable and closed population assumption but not robust to age-misreporting and differential under-registration of deaths by age after childhood |
Gray |
Assumes that the degree of completeness of death registration is more or less the same at all ages after infancy (or at age 1 or over) Assumes non-stable closed population |
Fairly robust to age-misreporting but not robust to differential under-registration of deaths by age after infancy or childhood |
Courbage and Fargues |
Assumes that the degree of completeness of death registration is more or less the same at all ages after infancy (or at age 1 or over) Assumes a model mortality pattern to represent the Philippine mortality pattern |
Fairly robust if the assumed model mortality pattern represents actual pattern in question Applicable to population greatly affected by migration Sensitive to age-misreporting and violation to constant under-registration of deaths by age after infancy or childhood |
The Courbage and Fargues and Preston and Coale procedures have additional requirements: a model age pattern of mortality and a provisional estimate of the rate of growth, respectively. Brass’s and Gray’s methods do not require knowledge of the growth rate. In fact, the rate of growth is one of their estimated outcomes. For instance, Gray’s approach makes full use of the implicit relationships between the sectional growth rates. Each procedure suggests use of alternative variants to come out with values to be later evaluated as to their reasonableness.
Given the tedious calculation required for each technique, the Small Working Group for the Mortality Estimation used two softwares. The PAS software contains the Brass Growth Balance and Preston and Coale methods each with two variants. The variants of the Brass Growth Balance method are the “mean” line (Line 1) and the “robust” line (Line 2) as basis of the selection of a best fitting line that yields the value of K, to derive the completeness of death registration (C= 1.0/K). The variants of the Preston and Coale method are the two types of ratios of estimated to reported population non-cumulative (N’(x,5)/N(x,5) and cumulative (N’(x to A)/N(x to A) as basis of calculating the median value which is the estimate of completeness of death registration.
The other software used was a modified Fortran program written by Dr. Alan Gray in 1986 and used by Dr. Cabigon for her Ph.D. dissertation. This program generates estimates using all four techniques of estimating completeness of death registration. However, it differs slightly from the PAS software for it calculates C (Completeness of death registration) based on the following:
Preparation of a short list of correction factors
The second step is to prepare a short list of correction factors (Cs) from the results in Step 1 to correct the registered deaths at ages 5 and over. It must be stressed here that owing to the unique assumptions and differing strengths and weaknesses between or among the four techniques, there will be no, or only occasional close correspondence between the results obtained using all of them. The likely situation, when all four show consistent levels of completeness of death registration, is when there are very few registered deaths, causing the application of all four techniques to reveal equivalent low values, which may be mainly statistical artifacts. In determining the extent to which these four methods are consistent with each other, the following conditions and the corresponding pair of combination were used as guide in the short listing:
| Condition | Pair of Combination |
Stable closed population with or without age-misreporting |
Brass’ and Preston and Coale’s |
Non-stable close population with or without age-misreporting |
Gray’s and Preston and Coale’s Gray’s and Brass’ Gray’s, Brass’s and Preston and Coale’s |
Appropriate model mortality pattern with: |
|
Stable open population with or without Age-misreporting |
Brass’s and Courbage and Fargues’ |
Non-stable open population with or Without age-misreporting |
Gray’s and Courbage and Fargues’ Preston and Coale’s and Courbage and Fargues’s |
Extremely poor registration |
A combination of all four methods |
One can establish many levels of consistency for any given pair of combination. However, the Small Working Group on Mortality took three arbitrary criteria as follows:
If close or moderate congruence is evident in at least one of the above eight pairs or a combination of all four, it may be taken that the Cs are reasonable. It must be stressed at this point that the main reason for using four different techniques is to select the technique that is robust for a particular aggregation. A given technique is only robust when its underlying assumptions are not strongly violated. The strategy adopted by the Small Working Group on Mortality in the final choice of the best level of completeness of death registration is not to opt uniformly for only one particular technique for all areal units under consideration. The main reasons are the great variations among regions and provinces in: (1) internal migration; (b) socioeconomic status; (c) mortality and fertility changes; and (d) quality of data over time.
Therefore, the results of the four approaches are generally used as guides to make a reasonable selection of the best level of completeness of death coverage for deaths ages 5 and over. Those in the resulting short list were further evaluated based on the following:
Generation of corrected deaths for ages 0 and 1-4
The third step is to generate the corrected deaths for ages 0 and 1 to 4 years. The short listed correction factors were used to correct the registered deaths for ages 5 years and over and to calculate the corresponding age-specific death rates (nmx) in the life table terminology. The resulting age specific death rates for the age group 5 to 9 (5m5) for each aggregation is used to derive the implied age specific death rates for ages 0 (1m0) and 1 to 4 (4m1) from the United Nations Latin American Model using the MORTPAK software.
Evaluation of the resulting age-specific death rates for ages 0 and over
The fourth step is to evaluate the resulting age-specific death rates for ages 0 and over as to their conformity with the death function, which is a U-shaped curved having its high point at age zero, then declining very rapidly to its minimum at around age 10 and thereafter increasing uniformly with age. The resulting age-specific death rates for ages 55 and over which deviated from this expected pattern were adjusted using the Gompertz curve.
Calculation and evaluation of life tables
The fifth step is to calculate the life tables after performing the above four steps.
The last step is to evaluate the resulting life tables based on the short list. These resulting life tables are further evaluated in accord with the following:
Only one assumption was used in projecting future changes in mortality using the life expectancy at birth as indicator. The projected values of life expectancy at birth for the five-year periods of the projection cycle were derived by applying the United Nations (UN) Model on quinquennial gains in life expectancy at birth following a moderate increase in survivorship (see table A). The resultant values are presented in Table 1.
Table A. Working Model for Mortality Improvement, Quinquennial Gains (in years) in Life Expectancy at Birth (e00) According to Initial Level of Mortality
Initial
mortality level |
Fast |
Middle |
Slow |
|||
Male |
Female |
Male |
Female |
Male |
Female |
|
|
|
|
|
|
|
|
55.0 - 57.5 |
2.5 |
2.5 |
2.5 |
2.5 |
2.0 |
2.0 |
57.5 - 60.0 |
2.5 |
2.5 |
2.5 |
2.5 |
2.0 |
2.0 |
60.0 - 62.5 |
2.5 |
2.5 |
2.3 |
2.5 |
2.0 |
2.0 |
62.5 - 65.0 |
2.3 |
2.5 |
2.0 |
2.5 |
2.0 |
2.0 |
65.0 - 67.5 |
2.0 |
2.5 |
1.5 |
2.3 |
1.5 |
2.0 |
67.5 - 70.0 |
1.5 |
2.3 |
1.2 |
2.0 |
1.0 |
1.5 |
70.0 - 72.5 |
1.2 |
2.0 |
1.0 |
1.5 |
0.8 |
1.2 |
72.5 - 75.0 |
1.0 |
1.5 |
0.8 |
1.2 |
0.5 |
1.0 |
75.0 - 77.5 |
0.8 |
1.2 |
0.5 |
1.0 |
0.3 |
0.8 |
77.5 - 80.0 |
0.5 |
1.0 |
0.4 |
0.8 |
0.3 |
0.5 |
80.0 - 82.5 |
0.5 |
0.8 |
0.4 |
0.5 |
0.3 |
0.3 |
82.5 - 85.0 |
- |
0.5 |
- |
0.4 |
- |
- |
85.0 - 87.5 |
- |
0.5 |
- |
0.4 |
- |
0.3 |
Note: Lower limit is inclusive of the value while upper limit is exclusive of the value. E.g. the range 65.0-67.5 means from 65.0 to less than 67.5.
Source: United Nations (1989), 1988 World Population Prospects, Table 1.4
Table 1. Projected Values of Life Expectancy at Birth by Sex, Philippines: 2000-2040
Period |
Life Expectancy at Birth |
|
Male |
Female |
|
2000-2005 |
64.10 |
70.10 |
2005-2010 |
66.10 |
71.60 |
2010-2015 |
67.60 |
73.10 |
2015-2020 |
68.80 |
74.30 |
2020-2025 |
70.00 |
75.50 |
2025-2030 |
71.00 |
76.50 |
2030-2035 |
72.00 |
77.50 |
2035-2040 |
73.00 |
78.30 |
This single set of projected values of life expectancy at birth, in combination with the three sets of fertility rates, were used to project the population up to the year 2040.
 |
  |
 |
||||||
  E-mail the webmaster |
Terms of Use |
Home • Top of Page | ||||||
1997-2008,
National Statistical Coordination Board |
||||||||