PRELIMINARY POPULATION HISTORY ESTIMATES OF CAMBODIA AND LAO PEOPLE'S DEMOCRATIC REPUBLIC

Transcription

1 PRELIMINARY POPULATION HISTORY ESTIMATES OF CAMBODIA AND LAO PEOPLE'S DEMOCRATIC REPUBLIC Feb Maks Banens Very little is known on the population history of Cambodia and Laos. The first modern, age and sex detailed censuses were held in 1962 (Cambodia) and 1985 (Laos). Until then, only one Demographic Sample Survey was organised in Cambodia French colonial statistics are not a great help either. They contain total population figures for 1911, 1921, 1931, 1936, 1946, and 1954 (Table 1). Not all of these were real census results. The 1931 figures result from local authorities evaluations rather then from a census : Les chiffres de population publiés ci-dessous ne résultent pas d un recensement véritable,..., mais représentent, d une manière générale, de simples évaluations provenant des déclarations effectuées par les autorités indigènes... (Annuaire statistique de l Indochine 1932, p. 49). The same method was used in 1911, 1936, and The latter census (1946) was, at least for Cambodia, especially under-enumerated because it had left to Siam part of its territory, notably the province of Battambang, re-integrated in Real censuses took place in 1921 and 1954 only. At the first census, its poor quality was admitted right a way by those who had carried it out. The Gouvernement Général de l Indochine observes for Cambodia : Les résultats du recensement serrent de près la réalité, sauf cependant dans les régions montagneuses de l Est, où la majeure partie des aborigènes, encore insoumis, ont échappé aux comptages et n ont été l objet d aucune évaluation numérique, et dans les villes... [où]... la population chinoise a été particulièrement sous-estimée. (Annuaire statistique de l Indochine 1927, p. 28) And for Laos : [Le recensement] présentait des difficultés particulières dues à la grande dispersion d une population très peu dense. Il semble toutefois que, sauf dans les régions insoumises de la chaîne annamitique et dans les montagnes du Haut-Laos, les chiffres indiquent bien l ordre de grandeur de la population ; on croit que, dans l ensemble, celle-ci excède de plus de 1% les chiffres officiels. (Ibid., p. 29) The second real census (1954) was disturbed by the Indochine War, when part of the territory was out of French control. Year (t) CAMBODIA total pop. (1) growth rate (t,t+n) % total pop. (1) LAOS growth rate (t,t+n) % Table 1. Total population and annuel growth rate until first modern census. Sources : Annuaire Statistique de l Indochine ; Heuveline 1998 ; Frisen Non-adjusted intercensal growth rates, as shown in table 1, may be biased upward as far as underenumeration tends to diminish over time. Especially the first growth rates are unlikely. However, from 1921 on, they might show something like the real trend. Unlike Vietnam, we have no estimate of vital rates over the colonial period. Actually, the first information comes with the modern censuses. For Cambodia, a Demographic Sample Survey took place a few years before the 1962 census, in It estimated life expectancy at birth at 44.2 for males and 43.3 for females (United Nations 1971, Demographic Yearbook 197, p. 72). The 1962 census, some years later, allowed to estimate life expectancies of 54.1 for males and 56.8 for females (Siampos 197, Heuveline 1998). For the early 199s, Banister and Johnson (1993) estimate a life expectancy for both sexes at 49 years only. For Heuveline (1998), a fifteenyear difference in life expectancy at birth between the two countries [Cambodia and Vietnam] appears large. (Heuveline 1998, p. 63) His medium

2 assumption for the early 199s holds 6. and 63. for male and female life expectancy at birth. Although our reconstruction covers the first half of the century only and stops in 1962, the uncertainties of the more recent period should warn us against too much pretensions. 7 6 life expectancy H assume that 2/3 of the emigration took place before the 1985 census, i.e. 27, net emigrants. Cambodia reconstructed Under the conditions mentioned above, using the three mortality scenarios of figure 1 and the West family of model life tables, I reconstructed the Cambodian population figures of the 1962 census. Figure 2 shows the ratios of reconstructed to enumerated male and female figures Reconstr/Enumer U.N. 1971, 1995 B.&J Figure 1. Life expectancy at birth, Cambodia both sexes ; recent estimates by Heuveline 1998, Banister and Johnson 1993, United Nations 1971 and Low, medium and high mortality reconstruction scenarios females males Mortality levels of Lao People s Democratic Republic are even less known. The first demographic survey started in It revealed a crude death rate of 1.6 % over the period , but no estimate of life expectancy was published. (Lao PDR 1989, Multiround Vital Statistics Survey : Preliminary results, July December 1988.) At the time of the first modern census, in Cambodia as in Laos, fertility was still at its pretransitional level. Heuveline 1998, assumes for Cambodia : that total fertility was around 6.7 before the census and not declining (p. 62). Frisen 1991, observes for Laos that : there is as yet no clear indication of a significant reduction in fertility (p. 59), the crude birth rate is estimated at 4.5 % for the period. International migration to and from Cambodia until 1962 has been considered as negligible for the reconstruction. The same assumption could not be made for Laos over the 197s and early 198s. Indeed, international migration flows have been important from the mid-7s on, all over Indochina. Frisen estimates Laos net emigration over 4, between 1975 and 1991 : The impact of this stream of emigration is evident today in the age-sex structure of the population (p. 62). Though he doesn t detail pre- and post-1985 migration, I will Figure 2. Cambodia. Ratio of reconstructed to enumerated 1962 census figures using high (red), medium (blue) or low (green) mortality scenarios. Sources : U.S. Bureau of Census, International Database. (-4, 5-9 and 1-14 age groups have been adjusted by U.S. Bureau of Census). All three mortality scenarios allow close reconstruction of the 1962 census. Correction is limited within 5 % until 55 year of age, and exceeds 1 % only for some open age group Older females may have been under-enumerated relatively more than older males. Or the type of mortality may have been different of the West family used here. At young ages, 1-14 females may have been under-enumerated while as 2-29 females seem over-enumerated. This could be explained by age misreporting : young females reporting older as they are. Yet, we should keep in mind that the 1-14 age group has been adjusted already. And, anyway, a 2 or 3 % correction is far too small to find rational explication, even in more developed countries. The 1962 reconstruction results cannot differentiate the three mortality scenarios. Total population and fertility results may do this. 2

3 Child/Female For the general fertility, the medium mortality scenario finds exactly the 6.7 value estimated by Heuveline for the last pre-census period. The high mortality scenario finds close to 7., the low scenario exceeds slightly 6.5. If the medium scenario seems optimal, differences are too small to be significant. On the other hand, the proximity of all estimated fertility indices to reported fertility is an argument for the robustness of the reconstruction procedure. Considering total population curves, I remind that the reported values for may have been too much disturbed to be taken into account. Over the previous period, we observe that the low mortality scenario gives the better fit. Yet, it would be hasty to conclude at its superiority. Reported total population figures before 194 are certainly under-estimated. We don t know by what factor, but considering fertility results, considering mortality estimates and scenarios, considering total population estimates and probably underenumerated reported total population, I personally privilege the medium mortality scenario Laos reconstructed 1985 Census results, as reported by Frisen, may have been adjusted without mentioning. Anyway, the age and sex distribution is very close to a stable population (West, 3 year life expectancy at birth, both sexes, r = 1.5%). Figure 5 shows stable population age and sex distribution together with the 1985 census figures, that were increased by 27, net emigrants (see text above), prorated over sex and age as the Vietnam emigrants estimated by Merli (1997, 1998). Figure 3. Cambodia. Estimated general fertility using high (red), medium (blue) and low (green) mortality scenarios. Coale s indice (I f ) in TFR equivalent (12.44*I f ). Total Pop thousands Figure 4. Cambodia. Reported and reconstructed total population using high (red), medium (blue) and low (green) mortality scenarios. Figure 5. Laos. Female (red) and male (blue) 1985 increased census results (dots) and stable population distribution (lines). See text for details. It strengthens the assumption that fertility did not yet fall in 1985, as Frisen points out. To reconstruct Laos population history, it then turns out more sensible to start from different fertility level assumptions. I used three scenarios. Medium fertility : CBR = 4.2 %, i.e. observed CBR in 1985 ; low fertility : CBR = 3.8 % ; high fertility : CBR = 4.6 %. Each scenario assumes constant fertility until 197, and graduate adjustment to observed level in Unlike Cambodia, the Laos reconstruction incorporates net migration estimates. Considering Laos comparable to Vietnam, I used Vietnam s net migration profiles as calculated by Merli (1997, 1998) to distribute total net emigration estimated at 27, (see above). 3

4 Reconstr./Reported Figure 6. Laos Ratio of female (dotted) and male reconstructed to reported figures using low (green) medium (blue) or high (red) fertility levels. Like Cambodia reconstruction, final 1985 fits hardly differentiate the three scenarios. Only the high fertility scenario shows more distance to reported males at older ages. Like Cambodia again, young adult males seem under-enumerated, even after net migration correction. And young female adults seem overenumerated. The correction factor is bigger as in the case of Cambodia, but it still remains reasonable considering the problematic reliability of the sources life exp. at birth, both sexes Figure 7. Laos. Estimated life expectancy at birth, both sexes, using low (green), medium (blue) or high (red) fertility scenarios ; estimated values (dots) and trends (thick lines) thousands Figure 8. Laos. Total population, reported (black) and estimated using high (red), medium (blue) or low (green) fertility scenarios. Implied life expectancies at birth for both sexes (West life tables) and estimated total population curves are showed in figures 7 and 8. Estimated life expectancy values are somewhat chaotic because they served as parameters to fit all 1985 age group figures and, thus, reflect every bump of the reported age and sex distribution. Yet, the trend over the period seems plausible, at least for high and medium fertility scenarios. The low fertility scenario implies life expectancies that seem too high too early. They are improbable, but not impossible. Total population curves all show much larger population estimates over the past. Twice to two and a half times official figures, whereas Cambodia s under-enumeration has been estimated at 1 to 4 %. I don t think migration movements could hold a substantial part of the explanation. I presume real and important under-enumeration of colonial statistics. The high fertility scenario implies a total population curve that is flat all over the first half of the century. This means no population growth at all. This might seem improbable. Medium and low fertility scenarios are more likely. Considering implied life expectancies, total population curves and 1985 census fits, I privilege the medium fertility scenario. 4

5 CITED REFERENCES Banister J. and E. P. Johnson (1993), "After the nightmare : the population of Cambodia", in B. Kiernan (ed.), Genocide and Democracy in Cambodia : the Khmer Rouge, the United Nations and the International Community, Monograph Series, 41. New Haven: Southeast Asia Studies, Yale University, pp Frisen C. M. (1991), "Population Characteristics in the Lao People's Democratic Republic", in Asia-Pacific Population Journal, Vol. 6, No. 2, pp Gouvernement Général de l'indochine (1927), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (193), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1932), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1933), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1935), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1937), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1938), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1939), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1942), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général de l'indochine (1948), Annuaire Statistique de l'indochine , Hanoi Gouvernement Général (1951), Annuaire Statistique de l'union Française Outre-Mer , Paris Gouvernement Général (1956), Annuaire Statistique de l'union Française , Paris Heuveline P. (1998), " 'Between one and three million' : Towards the demographic reconstruction of a decade of Cambodian history (197-79)", in Population Studies, 52, pp Lao PDR (1986), Population of Lao PDR, Vientiane Lao State Planning Committee (n.d.), Multiround Vital Statistics Survey: Preliminary results, July Dec Lewis L. (1991), "Multiround Vital Statistics Survey in the Lao PDR", in Asia-Pacific Population Journal, Vol. 6, No. 2, pp Merli G. (1997), Estimation of International Migration for Vietnam, , Seattle Population Research Center, Working Paper No. 97-4, 39 p. Merli G. (1998), "Mortality in Vietnam, ", in Demography, Vol. 35, No. 3, pp Siampos G. S. (197), "The population of Cambodia, ", in Milbank Memorial Fund Quaterly, 48, pp U.N. Development Program (1989), Development Co-operation: Lao PDR, 1988 Report APPENDIX 1 TABLE 1. CAMBODIA ESTIMATED FEMALES (medium mortality scenario)

6 TABLE 2. CAMBODIA ESTIMATED MALES (medium mortality scenario) TABLE 3. LAOS ESTIMATED FEMALES (medium fertility scenario) TABLE 4. LAOS ESTIMATED MALES (medium fertility scenario)