CAMBODIA: RECONSTRUCTING THE DEMOGRAPHIC STAB OF THE PAST AND FORECASTING THE DEMOGRAPHIC SCAR OF THE FUTURE

Transcription

1 CAMBODIA: RECONSTRUCTING THE DEMOGRAPHIC STAB OF THE PAST AND FORECASTING THE DEMOGRAPHIC SCAR OF THE FUTURE Ricardo F. NEUPERT Senior Researcher Center for Population Studies, Royal University of Phnom Penh United Nations Population Fund (UNFPA) Phnom Penh, Cambodia PRUM Virak Researcher Center for Population Studies, Royal University of Phnom Penh Phnom Penh, Cambodia Paper presented at the IUSSP Seminar on the Demography of Conflict and Violence, held in Oslo, Norway, 8-11 November The views expressed in this paper are those of the authors and do not necessarily reflect the views of the Royal Government of Cambodia, the Royal University of Phnom Penh or the United Nations Population Fund. The authors are indebted to the UNFPA, the Royal University of Phnom Penh and the National Institute of Statistics for their support and co-operation. We also would like to express our gratitude to Dr. Chiev Khus and Mr. Bjarke Oxlund for their many valuable suggestions and comments.

2 CAMBODIA: RECONSTRUCTING THE DEMOGRAPHIC STAB OF THE PAST AND FORECASTING THE DEMOGRAPHIC SCAR OF THE FUTURE After almost a century of being a French protectorate, Cambodia gained full independence under the leadership of Prince Norodom Sihanouk in November However, under his reign, internal political conflicts continued and in March 1970, a military coup led by General Lon Nol overthrew Prince Sihanouk. After five years of civil war and the spill over of the American-Vietnamese conflict into the country, on April 1975, the Khmer Rouge ousted the Lon Nol government and took control of the country. Just after taking power, the radical new regime forced the whole population of the capital city and provincial towns to leave for the countryside where they were placed in mobile teams and worked as forced labourers in the field for 12 or more hours a day. Cut off from the world, Cambodia come into a dark period, or year zero society, as all former national institutions were completely eradicated. On January 1979, Vietnam invaded Cambodia and, with the support of Cambodian anti-khmer Rouge forces, defeated the regime. In May 1993 free elections took place in the country, with a turnout of 89.6 percent and under the close supervision of the UNTAC (United Nations Transitional Authority in Cambodia). Since then the country was proclaimed as the Kingdom of Cambodia with a system of constitutional monarchy. At present, Cambodia is relatively stable political, economic and social terms (see, for example, Chandler, 1996). As it might be expected, the events of the 1970s resulted in a substantial increase in mortality, a decline in fertility and substantial emigration. These trends induced by the atypical political and economic situation experienced by the country, especially during the second half of the 1970s lasted only so long as that situation went on. With the normalization of the situation in the early 1980s fertility experienced a substantial increase that resulted in a typical post-war baby boom and then started an important decline. Mortality underwent a significant reduction, although it is still high in comparison with other Southeast Asian countries. The demographic consequences of the political and social turmoil of the 1970s form a clear indent in the population pyramid constructed from the most recent census (1998). Since there is no reliable registration, the volume of the losses is under discussion. Estimates go from one million to as high as three million direct and indirect deaths. There have been attempts to measure mortality using indirect methods and a voting census but the information has not been validated from other sources (Heuveline, 1994). Fertility and emigration have also been estimated, but only though the usage of unconventional or incomplete sources. The purpose of this study is to analyse the size and composition of the population of Cambodia in the past, present and future, especially in terms of the consequences of the traumatic events of the 1970s. A second purpose is to analyse how the population would have been at present and in the next two decades, if no conflict had taken place. In the first section of this study, the 1998 Census population is projected up to year The emphasis is on the analysis of the irregularities in the respective population pyramids caused by the 1970s events. In the second section the year 2000 population is retrospectively projected back to year 1980 and in the third section the population from the 1962 Census is projected up to year In the fourth section, the excess of deaths and the deficit of births are estimated for the decade by using the 1970 and 1980 populations and the estimated number of migrants. Finally, the 1970 population 2

3 is projected up to year 2020 under the assumption that the conflicts of the 1970s did not occur. The results of this projection are compared with those obtained by the ordinary projection. Therefore, this paper attempts to contribute to the reconstruction of the demographic history of Cambodia in three ways. First, by analysing scars in the pyramids corresponding to the recent past, present and future; second, by estimating major demographic events during the 1970s, especially excess deaths; and third, by comparing the size and composition of a simulated normal population with the real population. THE PRESENT AND FUTURE: The last population census in Cambodia was taken in Just after the results of the census were ready 1, the National Institute of Statistics (NIS) prepared population projections by age and sex at the national, provincial and rural-urban level (NIS, 1999). A standard cohort method was used for the projection. However, fertility in Cambodia substantially declined, especially during recent years. For example, according to the 2000 Cambodian Demographic and Health Survey (2000 CDHS) during the quinquenium , total fertility rate (TFR) was 4.0 children per woman as compared to 5.8 during and to the 1998 Census data estimate of (NIS, DGH and ORC Macro, 2001). It is considered that fertility estimates from the 2000 CDHS are more reliable than those estimated from census data (NIS-CPS, forthcoming). In other words, fertility decline was not well captured by the fertility estimates used in the projection (NIS, 1999). In addition, the application of other indirect methods to the 1998 Census information and other sources provided fertility and mortality estimates consistent with the 2000 CDHS (see PDPST, 2002 and MOP, 2002). Therefore, the NIS decided to update the previous projections using the new and more reliable estimates. The information presented in this section is from this updated projection 3. Figures 1a, 1b, and 1c show the population pyramids in absolute numbers corresponding to years 2000, 2010 and The most recent pyramid (year 2000) depicts a population with a young age composition with a typical triangular form, although with some major irregularities (the populations by age groups and sex on which all the pyramid in this paper were constructed are presented in Table 1 in the Appendix). FIGURES 1a TO 1c The first, most evident and general anomaly in the 2000 pyramid is the length of the bars starting in the age group years old (plain grey bars), which indicates a much smaller population than can be expected considering the bars corresponding to the population in age groups 0 to 19 years of age (patterned bars). The reason is that the age groups 20 and older were alive or were born during the 1970s and in particular during the Khmer Rouge period (April 1975 to January 1979). During this time, due to the political violence as well as harsh living and working conditions, mortality rates were extremely high. Fertility rates seem to have also been quite low. In addition, numerous people emigrated and many seem to have not returned. The second irregularity refers to the length of the bars corresponding to the population aged 40 and 74 (light grey). The number of males is sensible smaller than the number of females. During the Khmer Rouge period most of this population was 20 to 54 years of age. It could be suggested that adult males were more targeted by the political 3

4 violence than were women, children, teenagers and the elderly. Those age groups may have been affected by direct mortality caused by combat and political executions. Violent mortality certainly affected the whole population during the 1970s but it seems that the main victims were adult males 4. Initially, the main targets of the Khmer Rouge executions were personnel from the Lon Nol regime and later on suspected traitors within the party (Heuveline, 1998). Harsh living conditions and lack of health services may have also affected the mortality of these age groups but direct mortality causes appears to have play a major role in the length of the respective pyramid bars. Notice, however, that in the year 2000 pyramid the male and female bars corresponding to the age groups from 20 to 39 years old are similar in length (dark grey). Actually among these age groups there are more females than males but the difference is not as large as in the upper age groups (see Table 1 in the Appendix). These age groups were born during the 1970s or were younger than 20 years old. The excess mortality seems to have affected similarly both sexes among these age groups. Indirect mortality associated with poor nutrition and a deterioration of health conditions may have affected the survival of this segment of the population. These indirect mortality causes does not seem to discriminate very much between sexes. They may also have experienced direct mortality, especially those older than 15 years of age, but in general mortality among these age groups seems to have been caused mainly by indirect causes. It is also important to notice that the age groups born during 1970s (20-24 and years old in year 2000) has a size similar to that born during the previous decade ( ) and are and years old in year Considering the level of fertility during the 1960s (TFR=6.85), it seems that TFR substantially declined during the 1970s. Assuming that the mortality conditions affected equally the four age groups under consideration, one might have seen longer bars for the age groups and if fertility would have remained constant or experienced only a minor decline during the 1970s. However, the respective bars are not much longer that the previous two (30-34 and years old, that were born during the 1960s). This pattern suggests an important fertility decline during the 1970 and will be analyzed further in the following sections. The three age groups corresponding to the population aged 5 to 19 (checked bars in the pyramid) were born after the 1970s, that is, between 1980 and The pattern suggests that after the Khmer Rouge period a baby boom took place (Desbarats, 1995, Dasvarma and Neupert, 2002). Considering the length of the bars corresponding to the age groups from 20 to 39 years in the 2000 pyramid, the larger size of the groups from 5 to 19 appear to be the result of both mortality decline and an increase in fertility. Actually, during the period from the early 1980s to the mid-1990s fertility seems to have declined slightly as compared to pre-conflict levels, but it was much higher than during the 1970s. As mentioned in endnote 3, according to estimates based on the 2000 CDHS, TFR was 6.1, 5.9 and 5.8 in , and , respectively (NIS, DGH and ORC Macro, 2001). This information confirms that a post conflict baby boom occurred. As pointed out before, starting in the second half of the 1990s fertility substantially declined: 3.99 for the period This decline is quite clear in the year 2000 pyramid. The age group 0-4 bars (grey bars with vertical lines in the pyramid) are shorter than the previous bars corresponding to the group 5-9. Between and TFR declined by 31%. 4

5 The year 2010 pyramid (Figure 1b) shows the 2000 pyramid's age groups 10 years later. The segments that were born after 1980 (patterned bars) are replacing those that experienced the hardship and violence of the decade (plain grey bars). However, the latter segment will leave a mark in the former. Notice that the bars for the age group 0-4 in the 2010 pyramid (bars with diagonal lines) are longer than those corresponding to the two older age groups. These bars, corresponding to the age groups and 5-9 (grey bars with lines), are mainly the children of the 1960s and 1970s generations (dark grey bars) but the age group 0-4 are the children of the age groups born after the 1970s (checked pattern bars). Fertility is assumed to decline among those groups in the projection; the assumption, as mentioned in footnote 3, is that it will continue to fall during the complete projection period. The reason for the length of the 0-4 years old bars is that they represent mainly the children of the population born after 1980 (checked pattern bars), which is larger than the segments born during the 1960's and 1970s (because of lower mortality). Thus, large the post-1980 age groups gave birth to a larger number of children than the smaller generations born in the 1960s and 1970s that were affected by the mortality excess and low fertility. In other words, the comparatively larger bar representing the 0-4 years old population is the result of a population momentum. This is even more evident in the 2020 pyramid (Figure 1c). The children born from the 1960s and 1970s generations are represented by shorter bars (grey bars with vertical lines) producing a sort of indent in the pyramid. The first three bars are the children of the baby boom cohort born (bars with diagonal lines). Finally, it is worth to state that the annual rate of population growth for the decade is lower than that corresponding to (Figures 1b and 1c) in spite of the fact that in the projection fertility is assumed decline. The reason is the population momentum provided by the comparatively larger age groups that were born after the Khmer Rouge period ( ) under a demographic regime of relatively high fertility and a sudden improvement in survival probabilities. THE RECENT PAST: The 2000 Cambodian Demographic and Health Survey (2000 CDHS) included questions directed to construct birth histories and, therefore, obtain fertility and early-age mortality rates for about 20 years back (NIS, DGH and ORC Macro, 2001). Using that information and the year 2000 population by age groups and sex (obtained from the population projection used in the previous section; see endnote 3) a retrospective projection was conducted in order to estimate the population from 2000 back to In other words, the year 2000 population was reverse projected back to Figures 2a to 2c show the pyramids corresponding to the estimated populations of 1980, 1990 and Notice that for comparative purposes they have the same scale as the previous pyramids (years 2000, 2010 and 2020). Table 1 in the Appendix shows the respective numbers. It is important to point out that other authors have also estimated the population of For example, Huguet (1997) used the number of voters registered by the United Nations Transitional Authority in Cambodia (UNTAC) by the end of 1992, the 1996 Demographic Survey of Cambodia (NIS, 1996), and employed several indirect demographic techniques and a reverse projection method. His estimate of the total 5

6 population for mid-1980 was 6.55 million. According to Table 1 in the Appendix, the population estimated in this exercise was 6.80 million. The result obtained here is somewhat larger because Huguet used lower life expectancies (that is, higher mortality), lower fertility rates but mainly because he estimated a larger number of return migrants for the period. In 1980, the Government of the People's Republic of Kampuchea conducted a population count with a result of 6.56 million (cited by Huguet, 1997 and Heuveline, 1998). The United States Bureau of the Census (1983) estimated a population of only 5.7 million for the same year and the United Nations (1995) provided a figure of 6.7 million. The advantage of the population size and composition estimated in the present study is that they were obtained mainly with conventional demographic sources (namely the 1998 Census and the 2000 CDHS), which can be considered more reliable than unconventional sources such as electoral data or fragmentary registers. FIGURES 2a TO 2c The pyramid 2a and 2b, representing the 1980 and 1990 populations are quite consistent with the year 2000 pyramid. Notice that the 1980 pyramid the bars, representing the age groups from 0 to 19 (dark grey), which were children or were born during the 1970s, exhibit a similar number of male and females; also the size of the bars are similar. Among the population 20 to 54 (light grey) there is a larger number of females. The explanation of there trends was given in the previous section when the projected 2000, 2010 and 2020 pyramids were examined, and these sectors were much older. It is important, however, to emphasise the length of the bars corresponding to the age groups from 0 and 19 years old in the 1980 pyramid. As mentioned above, the fact that the respective bars are not arranged in a set of ascending steps with a clear slope suggests a decline in fertility during the 1970s. Notice that in the pyramid representing the 1990 population the two age groups born after the Khmer Rouge period emerged (checked pattern bars) and are represented by a much longer bar that the age groups over them. As pointed out above, these are the baby boom age groups; in addition, their members did not experience the conditions that resulted in the excess mortality of the previous decade. The year 2000 pyramid was already analysed and was included here to help to visualise the consistency of the backward projection. However, it is worth to repeat the appearance of the age group (0-5 years old) corresponding to the substantial fertility decline experienced by the Cambodian population during the second half of the 1990s (grey bars with vertical lines). THE PRE-CONFLICT PERIOD: The census conducted before 1998 was done in The 1965 and 1970 populations were estimated based on this census 6. The pyramids representing these populations are shown in Figures 3a and 3b; the respective numbers are in Table 1 in the Appendix. Notice that the age groups that were examined in the previous pyramid begin to appear in these figures. FIGURES 3a AND 3b It should be mentioned that these pyramids are normal, that is, they have the regular triangular shape of a young population with high fertility and mortality rates, with no important irregularities. It is also relevant to notice that the differences between males and females are much smaller than in the pyramids corresponding to 1980 and forward. For 6

7 example, in 1970 the sex ratio is men per 100 women, but in 1980 it is In year 2020, according to the projection, the sex ratio will return to normal with a value of In 1970, for the age group years, the sex ratio is 97.3 males per 1000 females but in year 1980 the ratio is For the same years the sex ratios for the population 0 to 19 years old are and These data confirm that during the 1970s, there was an excess of male deaths, especially among the adult population. Among children and teenagers, the difference is much smaller. As suggested above, this is likely to be related to the fact that adult deaths were predominately caused by violence, while children and adolescents deaths were related indirect causes such as poor nutrition and lack of health services, which do not discriminate by sex. THE TRAGIC DECADE: Until now, the populations corresponding to the periods and have been projected forward or backwards. In this section an attempt is made to estimate the excess deaths and deficit of birth during the period. For this purpose the 1970 population was projected up to 1980 using a standard cohort component method, assuming a normal mortality and fertility. The result provides an expected population, that is, the population that would have survived under normal conditions (without any conflict) 7. If the 1980 observed population (reverse projected) is subtracted from the expected population, the result will be the population that died, emigrated or was not born because of the conflict. In other words, the difference between the two populations is the result of emigration, excess deaths and deficit of births that affected the observed population during the 1970s. In order to know the excess number of deaths it is necessary to subtract from this difference the number of emigrants and the number of births. The number of emigrants was obtained from Heuveline (1998). Based on diverse estimates, in his study on mortality during the 1970s in Cambodia, he proposes three hypotheses regarding migration. In his medium hypothesis he recommends a net migration of 350,000 persons. This figure was used in the present analysis. It was dieseaggregated by sex and age groups assuming that migrants had the same sex and age composition of the country's population at the time. To estimate the number of births, a procedure, based on the fact that the first four bars of the 1980 pyramid do not exhibit the typical pyramid shape of upward decreasing bars, was used. The strategy was as follows. The age group 5-9 years old in 1980 was increased by the ratio between age groups and 5-9 years old in 1970 and the age group 0-4 years old in 1980 was increased by the ratio between 5-9 and 0-4 years old in This procedure equalizes the age ratios of age groups 5-9 and 0-4 between years 1970 and 1980 and also equalizes the age ratios of the age groups and 5-9 between the same years. This technique was applied to each sex independently. It can be assumed that the amount of children needs to equalize age groups 0-4 and 5-9 corresponds to the deficit in births during the 1970s. This is illustrated in Figure 4. The grey portion of the pyramid corresponds to the observed population in 1980 and the white portions of the first two bars relate to the deficit of birth. The inclusion of these portions make the four first bars of the pyramid appear with a normal shape, that is, with a regular 7

8 slope and with the bars forming stairs of approximately the same size. It is important to point out that this is not an accurate method to estimate the deficit in the number of birth resulting from the fertility decline that appears to have occurred during the 1970s, but other sources are not available. In addition, the estimate appears to be acceptable for the purposes of this paper. FIGURE 4 The migrants and the number of births were subtracted to the difference between expected and observed population. The stages of the calculation and the results are presented in Table 1. According to this estimate, the number of excess deaths during the 1970s was 2.27 million and the deficit of birth about 371 thousand. TABLE 1 Several quantifications of the dramatic mortality of the 1970s are available. The approaches have been twofold. The first one begins with a reverse count of deaths in a sample obtained from grave excavations or survivors' oral communications. The sample number is then extrapolated to the population of the country. These estimates have been from 1.5 million and 2 million. The reliability of these calculations cannot be reasonably evaluated because of the limited information on the population being sampled. Also, the data are difficult to verify. The second approach is based on demographic analysis. The balance equation is a basic demographic formula to calculate change in population size between two dates. This change is the result of the number of births, deaths and net migrants between the two dates. Therefore, in a given period excess mortality can be estimated as a residual in the balancing equation of population change between two dates. This demographic approach has produced lower mortality estimates, from 0.7 to 1.2 million (Heuveline, 1998; see his paper for the source of the number of deaths previously mentioned). The most systematic and careful example of this approach was done by Heuveline (1998). The tactic used in the present paper is similar to the one he employed in his study. However, his estimates are only for the population born before 1970 (he does not include early-age mortality). Under different assumption for mortality and migration he obtained three estimates: 1.17 million (minimum), 2.52 million (medium) and 3.42 million (maximum). In the present study, the population 10 years and older is 1.48 million. Demographic reconstruction is hindered by the reliability of measurement in each population component, which are compounded in the estimate of the residual component. One of the main advantages of the present estimates is that the information to estimate the 1980 population is reasonably reliable (the 1998 Census and the 2000 CDHS). It is unlikely that this information can be improved to obtain more precise estimates of the 1970s. As time goes by, the quality of past information provided by interviewees will diminish. However, the demographic traces of the decade of the 1970s will continue being present in the composition of the Cambodian population. Irrespective of the uncertainties of the demographic reconstructions carried out about Cambodia, all of them reveal the tragic consequences of the events that took place during the 1970s. It is also important to point out that excess deaths is only part of the burden of the decade and, in particular, the Khmer Rouge period. Non-returning migrants, a birth deficit, and age composition distortions complete the present and future legacy of those four years. It is relevant to clarify that the present analysis is not finished. There are two analyses that were not completed in this first version: a classification of the 1970s deaths in terms of violent deaths (combat, executions) and non-violent deaths (deterioration of 8

9 health conditions, nutritional problems); another important exercise is to locate the deaths throughout each year, or at least each quinquennium of the decade. The work in this direction is in progress SIMULATON The final exercise is directed to assess the population size and composition up to year 2020 if the tragic events of the 1970s would have not occurred. The normal population projected for 1980 was projected forward up to year 2010 and compared with the observed population obtained in the first projection ( , see endnote 3) and the one obtained in the backward projection ( , see endnote 5). The results are presented in Figures 5a to 5e and also in Table 2 FIGURES 5a TO 5e TABLE 2 Each figure shows two superimposed pyramids. The grey one corresponds to the observed population (projected considering the demographic effects of the 1970s) and the white one is the expected population (the population under a situation of no conflict in the 1970s). Notice that the white pyramids exhibit the quite regular triangular shape of a population with high fertility and mortality. However, in year 2000 its base became smaller than the previous bars. This is because in the expected populations the decline in fertility that took place during the second half of the 1990s was included. In the years 2010 and 2020, after the indent caused by the decline in fertility, the shape of the pyramid become regular again. In spite of the assumption of fertility decline, the child population continues to grow because of the past large cohorts (momentum). Contrasting with the regular shape of the pyramids corresponding to the expected population, the observed pyramids (grey) show the scars of the 1970s, which will be visible for a long time. 9

10 10

11 REFERENCES Arriaga, E. (1994) Population Analysis with Microcomputers (PAS), U. S. Bureau of the Census, USAID, UNFPA, Washington, D. C., USA. Candler, D (1996) A History of Cambodia (second edition) Westview Press Inc. Boulder, Colorado, USA. CELADE (Latin American Demographic Center) (1992) PRODEM, Version 2, CELADE, Santiago, Chile. Dasvarma, G. and R. Neupert (2002) Fertility Trends in Cambodia, Paper presented at the 2002 IUSSP Regional Population Conference held in Bangkok, Thailand, June, Desbarats, J. (1995) Prolific Survivors: Population Change in Cambodia, , Arizona State University, Tempe, Arizona, USA. Heuveline, J. H. (1998) 'Between one and three million: Towards the demographic reconstruction of a decade of Cambodian history ( )'. Population Studies, 52, Huguet, J. W. (1997) The Population of Cambodia, , and projected to 2020, National Institute of Statistics, Phnom Penh, Cambodia. MOP (Ministry of Planning) (2002) Fertility and Family Planning in Cambodia, Ministry of Planning, Phnom Penh, Cambodia. NIS (National Institute of Statistics) (1999) Population Projections Report 6, National Institute of Statistics, Ministry of Planning, Phnom Penh, Cambodia. NIS (National Institute of Statistics) (1996) Demographic Survey of Cambodia General Report, National Institute of Statistics, Phnom Penh, Cambodia. NIS (National Institute of Statistics), CPS (Center for Population Studies) (forthcoming) Population Projections for Cambodia: Updated. National Institute of Statistics and Center for Population Studies, Royal University of Phnom Penh, Phnom Penh, Cambodia. 11

12 NIS (National Institute of Statistics), DGH (Directorate General for Health) and ORC Macro (2001) Cambodian Demographic and Health Survey, NIS, DGH, ORC Macro, Phnom Penh, Cambodia and Calverton, Maryland, USA. PDPST (Population and Development Policy Support Team) (2002) 'Mortality in Cambodia: A technical note', Technical Note # 1, Population and Development Policy Support Team, Ministry of Planning, July Siampos, G. S. (1970) 'The population of Cambodia, ', Milbank Memorial Fund Quarterly 48, pp United Nations (1883) Manual X. Indirect Techniques for Demographic Estimation, United Nations, New York, USA. United Nations (1988) MortPak-The United Nations Software Package for Mortality Measurement, United Nations, New York, USA. United Nations (1995) World Population Prospects: The 1994 Revision, United Nations, New York, USA. United States Bureau of the Census (1983) World Populations, 1983: Recent Demographic Estimates for the Countries and Regions of the World, United States Government Printing Office, Washington, D.C., USA. 12

13 FIGURES AND TABLES Figure 1a: Population Pyramid of Cambodia, year 2000 (absolute numbers) Male Female Figure 1b: Population Pyramid of Cambodia, year 2010 (absolute numbers) Male Female Figure 1c: Population Pyramid of Cambodia, year 2020 (absolute numbers) Male Female

14 Figure 2a: Population Pyramid of Cambodia, year 1980 (absolute numbers) Male Female Figure 2b: Population Pyramid of Cambodia, year 1990 (absolute numbers) Male Female Figure 2c: Population Pyramid of Cambodia, year 2000 (absolute numbers) Male Female

15 Figure 3a: Population Pyramid of Cambodia, year 1965 (absolute numbers) Male Female Figure 3b: Population Pyramid of Cambodia, year 1970 (absolute numbers) Male Female

16 Figure 4: Population Pyramids of Cambodia, estimated population and expected number of births, year 1980 Male Female

17 17

18 Table 1: Estimate of the deficit of births and the excess of deaths, Age 1980 expected population 1980 observed population Expected population minus (Projected from the 1962 Census) (Backward projection) observed population Male Female Total Male Female Total Male Female Total , ,782 1,722, , ,108 1,018, , , , , ,316 1,450, , , , , , , , ,667 1,236, , , , , , , , ,500 1,059, , , , , , , , , , , , , ,313 76, , , , , , , , ,335 82, , , , , , , , ,711 54, , , , , , , ,920 62,787 34,912 97, , , , , , ,912 57,957 41,116 99, , , , , , ,399 50,189 38,488 88, , , ,969 98, , ,495 41,145 29,329 70, , , ,529 82,817 96, ,866 26,532 21,131 47, ,155 92, ,208 71,150 79, ,778 13,005 12,425 25, ,242 71, ,850 49,695 58, ,079 14,547 13,224 27, ,890 47,929 89,819 31,041 36,425 67,466 10,849 11,504 22, ,982 45,613 82,595 28,627 35,442 64,069 8,355 10,171 18,526 Total 4,904,365 4,892,234 9,796,599 3,163,003 3,640,165 6,803,168 1,741,362 1,252,069 2,993,431

19 Table 1, continued Age Observed 1980 population minus Estimated number of Net migration Estimated deficit of births net migration and deficit of births excess deaths Male Female Total Male Female Total Male Female Total Male Female Total ,994-30,888-62,882 91,014 98, , , ,543 1,081, , , , ,034-25,978-53,012 91,238 90, , , , , , , , ,228-21,629-43, , , , , , , ,127-18,870-37, , , , ,966 82, , ,425-14,380-28, , , , ,888 61, , ,863-11,197-22, , , , ,472 71, , ,680-10,210-19, , , ,123 92,031 43, , ,700-9,184-17, , , ,805 54,087 25,728 79, ,658-7,778-15, , , ,348 50,299 33,338 83, ,208-6,360-12, , , ,966 43,981 32,128 76, ,057-5,228-10,286 93, , ,781 36,088 24,101 60, ,234-4,397-8,631 78,583 91, ,497 22,298 16,734 39, ,220-3,363-6,583 67,930 76, ,361 9,785 9,062 18, ,298-2,425-4,723 47,397 55, ,802 12,249 10,799 23, ,422-1,532-2,954 29,619 34,893 70,420 9,427 9,972 19, ,114-1,319-2,433 27,513 34,123 66,502 7,241 8,852 16,093 Total -175, , , , , ,197 2,805,488 3,276,482 7,153,169 1,383, ,386 2,272,233 19

20 Figure 5a: Population Pyramids of Cambodia, projected (gray) and simulated (white), year Male Female Figure 5b: Population Pyramids of Cambodia, projected (gray) and simulated (white), year Male Female Figure 5c: Population Pyramids of Cambodia, projected (gray) and simulated (white), year Male Female

21 Figure 5d: Population Pyramids of Cambodia, projected (gray) and simulated (white), year Male Female Figure 5e: Population Pyramids of Cambodia, projected (gray) and simulated (white), year Male Female

22 Table 2: Projected and simulated population of Cambodia, 1980 to 2020 Age Simulated Projected Simulated Projected Male Female Total Male Female Total Male Female Total Male Female Total , ,782 1,722, , ,108 1,018,194 1,153,514 1,111,861 2,265, , ,363 1,803, , ,316 1,450, , , , , ,190 1,930, , ,180 1,576, , ,667 1,236, , , , , ,022 1,647, , , , , ,500 1,059, , , , , ,624 1,404, , , , , , , , , , , ,649 1,189, , , , , , , , , , , ,716 1,010, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,969 98, , , , , , , , , , , ,529 82,817 96, , , , ,170 97, , , ,155 92, ,208 71,150 79, , , , ,852 81, , , ,242 71, ,850 49,695 58, ,079 81,827 94, ,971 62,278 78, , ,890 47,929 89,819 31,041 36,425 67,466 54,495 65, ,972 46,330 57, , ,982 45,613 82,595 28,627 35,442 64,069 54,657 72, ,681 40,705 57,327 98,032 Total 4,904,365 4,892,234 9,796,599 3,163,003 3,640,165 6,803,168 6,613,390 6,591,674 13,205,064 4,605,832 5,047,792 9,653,624

23 Table 2, continued Age Simulated Projected Simulated Projected Male Female Total Male Female Total Male Female Total Male Female Total 0-4 1,107,591 1,097,957 2,205, , ,139 1,719,671 1,280,307 1,241,450 2,521, , ,865 1,925, ,223,678 1,221,739 2,445, , ,274 1,901,630 1,134,169 1,107,412 2,241, , ,526 1,659, ,100,788 1,072,461 2,173, , ,559 1,744,740 1,067,997 1,063,871 2,131, , ,697 1,656, , ,714 1,870, , ,677 1,540,633 1,188,105 1,195,831 2,383, , ,476 1,850, , ,705 1,585, , , ,426 1,059,142 1,047,623 2,106, , ,113 1,689, , ,462 1,339, , , , , ,370 1,799, , ,407 1,481, , ,656 1,127, , , , , ,107 1,517, , , , , , , , , , , ,061 1,275, , , , , , , , , , , ,284 1,064, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,437 92, , , , , , , , , , , ,790 72, , , , , ,593 88, , , ,556 91, ,423 51,834 73, ,308 88, , ,965 61,348 93, , , , ,791 52,384 79, ,829 96, , ,129 80, , ,428 Total 8,343,354 8,456,330 16,799,684 6,078,533 6,495,047 12,573,580 10,093,183 10,256,538 20,349,721 7,436,843 7,831,745 15,268,588 23

24 Table 2, continued Age 2020 Simulated Projected Male Female Total Male Female Total 0-4 1,480,524 1,424,570 2,905,094 1,155,322 1,105,796 2,261, ,389,220 1,343,359 2,732,579 1,077,017 1,033,854 2,110, ,255,663 1,221,491 2,477, , ,817 1,883, ,114,673 1,095,175 2,209, , ,958 1,632, ,041,617 1,050,246 2,091, , ,007 1,621, ,151,042 1,176,738 2,327, , ,236 1,801, ,022,842 1,027,345 2,050, , ,173 1,639, , ,045 1,744, , ,928 1,431, , ,428 1,460, , , , , ,853 1,214, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,247 87, , , , , , , , ,299 Total 12,303,836 12,491,642 24,795,478 9,192,926 9,531,389 18,724,315 24

25 APPENDIX Table 1: Estimated and projected population of Cambodia by age groups and sex, 1965 to Age Male Female Total Male Female Total Male Female Total , ,541 1,149, , ,316 1,309, , ,108 1,018, , , , , ,058 1,104, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,161 98, , , , ,223 98, , , ,531 77, ,159 88,190 91, ,770 82,817 96, , ,952 58, ,932 67,067 70, ,108 71,150 79, , ,476 40,603 80,079 47,865 50,498 98,363 49,695 58, , ,232 24,692 47,924 29,617 31,907 61,524 31,041 36,425 67, ,312 21,846 40,158 23,196 27,454 50,650 28,627 35,442 64,069 Total 3,140,748 3,137,154 6,277,902 3,650,242 3,639,303 7,289,545 3,163,003 3,640,165 6,803,168 25

26 Table 1(Appendix), continued Age Male Female Total Male Female Total Male Female Total Male Female Total , ,363 1,803, , ,139 1,719, , ,865 1,925,775 1,155,322 1,105,796 2,261, , ,180 1,576, , ,274 1,901, , ,526 1,659,878 1,077,017 1,033,854 2,110, , , , , ,559 1,744, , ,697 1,656, , ,817 1,883, , , , , ,677 1,540, , ,476 1,850, , ,958 1,632, , , , , , , , ,113 1,689, , ,007 1,621, , , , , , , , ,407 1,481, , ,236 1,801, , , , , , , , , , , ,173 1,639, , , , , , , , , , , ,928 1,431, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,321 92, , , , , , , , , ,278 78, ,894 72, , ,271 88, , , , , , ,330 57, ,219 51,834 73, ,308 61,348 93, ,410 87, , , ,705 57,327 98,032 52,384 79, ,829 80, , , , , ,299 Total 4,605,832 5,047,792 9,653,624 6,078,533 6,495,047 12,573,580 7,436,843 7,831,745 15,268,588 9,192,926 9,531,389 18,724,315 26

27 ENDNOTES 1 It is important to mention that the enumerated 1998 census population was adjusted according to several assessments (see NIS, 1999 and NIS-CPS, forthcoming): (a) A post-enumeration survey was conducted after the census. It was estimated that the net census undercount was 1.78% of the total population. Thus the total population size was increased by this percentage (or divided by ). This adjustment was applied equally to each sex and age (b) At the time of the census (March 1998) it was not possible to enumerate the population of four small areas of the country because of security reasons. These areas were the district of Antong Veaeng in Otdar Mean Province, The village of Ou Bei Choan in Banteay Mean Chey Province, the district of Samlot in Bat Dambag Province, and the district of Veal Veaeng in Pousat Province. Exact estimates of the population in these areas are not available, but it was estimated that their total population was 45,000 persons. For the national projection, this number was added to the census population and it was done according to the age and sex distribution of the national population (c) In March 1998 there were 60,000 Cambodians temporarily displaced in camps in Thailand due to conflicts near their villages. Since all of them were supposed to have returned to Cambodia within few months after the census, they were included in the population of the country for projection purposes. It was assumed that 30,000 returned to Banteay Mean Chey Province and 30,000 to Bat Dambang Province. Hence the national population was increased by 60,000 and it was assumed that the returnees had the same age and sex distribution as the national population. (d) In most censuses problems of under-enumeration and age misreporting for the age group 0-4 are quite common. In order to correct data for this age group the following procedure was carried out in the original projections. It was assumed that fertility had been constant for five years prior to the census, with a TFR of 4.0. The 2000 Cambodian Demographic and Health Survey (2000 CDHS) estimated such rate. The census female population was reverse-projected by single years. The agespecific fertility rates for were applied to the backward projection of women in order to obtain the number of births in each of the five years prior to the census. Those numbers of birth were then survived to the census date using a mortality level estimates also from the 2000 CDHS. These calculations suggest that the number of males, aged 0-4 years, enumerated by the census should be increased by a factor of and the number of females aged 0-4 should be increased by a factor of These adjustment fac tors were applied to the census population (as adjusted for the three types of under enumeration mentioned above). 2 This estimate was obtained using a Brass-type indirect method (NIS, 1999) 3 For the new projection the base population was the adjusted 1998 Census (see endnote 1). Regarding the estimate of mortality, a model life table was constructed using infant mortality data from the 2000 CDHS; adult mortality was computed by applying indirect methods to the 1996 Demographic Survey of Cambodia. Adjustments were done using the United Nations software MORTPAK (1988). The level of mortality, in term of life expectancy at birth, for , was estimated to be 53.1 years for males and 59.2 years for females. Mortality levels were projected assuming that in year 2020 will be as those observed in Phnom Penh in 1998, that is, 68.4 years for males and 74.8 years for females. Life expectancies for each quinquenium between the base and the final year of the projection period were estimated linearly. The life tables corresponding to the projected life expectancies were produced using the software MORTPAK previously cited. The 2000 CDHS estimates of fertility were used (TFR=3.99 for ). Using 1998 Census data, fertility was also estimated. The Brass, Rele, and Own-children methods were

28 employed. The results were consistent with the survey estimates. The results indicate that the fertility of the population of Cambodia is experiencing a substantial decline. As mentioned previously, during the 1960's TFR was over 6 children per woman and by the late 1990s about 4 children. Using the available data several functions were utilized to extrapolate the observed trend in TFR. Finally, a logistic curve was considered to be the most appropriate. This function to project TFR requires four values: an upper asymptote, that should indicate the historical high fertility levels; a lower asymptote, that should indicate a level of fertility expected for the distant future, usually the level corresponding to replacement (a TFR of 2.1); and two intermediate TFRs with values between the upper and lower asymptotes. In the present projections, a TFR of 6.85 children per woman was used as the upper asymptote and 2.10 was used as the lower asymptote. The upper asymptote was estimated as follows. Using the 1962 Census, and the Brass-type indirect method, a TFR of 6.85 was obtained. This estimate of TFR corresponds to the five years before the census, which was , or to Since the analysis of the results of the application of the Brass method suggest that fertility has not declined it was assumed that 6.85 is the historical fertility level of Cambodia and, therefore, it is used as the upper asymptote. As a lower asymptote the replacement level fertility will be used (2.1). This is a common practice in this type of fertility projections. The TFR of 3.99 estimated by the 2000 CDHS for 1998 was used as the first intermediate value of the logistic. As the second intermediate value, a TFR of 3 children per woman in year 2020 was proposed. This value was assumed based on two considerations. First, according to the 2000 CDHS, the ideal number of children reported by women who are beginning their reproductive carriers (aged years) is 3.1. Second, also according to the 2000 CDHS, women with secondary education or more had a TFR of 2.9 children in In many countries it has been frequent an overall declining fertility trend toward the fertility observed among the most educated women. Therefore, it is reasonable to expect a TFR of 3.0 by year The program LOGISTIC from the package PAS (Arriaga, 1994) was used to adjust the logistic curve to the parameters mentioned above. The 1998 Census included three questions regarding internal and international migration; however, regarding international migration, it is possible to estimate only immigration. It seems unlikely that international migration will become a major component of population change in the country. This is not to say that migration will not take place in the future 20 years, but probably its level will be low and hence unlikely to impact on the population size and structure. This is only an assumption and does not mean that in the near future emigration or immigration will not occur; however there is no evidence for proposing a hypothesis regarding volume and trend of international migration. The usual practice in this case it is to assume that it will be negligible. Finally, it is relevant to mention that the software used for this projection was RUP, developed by the US Bureau of the Census (see Arriaga, 1994). 4 In most countries, among the population aged 60 and over is frequent to observe a large number of females than males. Therefore, in this case, the differences over age 60 in year 2000 is probably not only the result of an excess of male mortality during the Khmer Rouge period but the higher life expectancy of females observed in most places. 5 The method used for the retrospective projection was a cohort component method. As mentioned, the base population was the projected year 2000 population that was backward projected toward year Mortality was calculated from the infant mortality estimated by the 2000 CDHS, which were 84.6, 78.8, 91.1 and 95.0 for the quinquenia , , and , respectively (NIS, DGH and ORC Macro, 2001). Using the program MORTPAK (United Nations, 1988) and the model life table estimated in the previous projection (see note 3), life expectancies at births and life tables were obtained. Life expectancies were 55.1, 56.6, 53.5 and 52.6 years for males and 61.4, 62.9, 59.7 and 69.6 years for females for the quinquenia , , and , respectively. 28