LEVELS AND TRENDS OF FERTILITY AND MORTALITY IN BURMA

Transcription

1 LEVELS AND TRENDS OF FERTILITY AND MORTALITY IN BURMA by Nyan Myint A thesis submitted in partial fulfilment of requirements for the Degree of Master of Arts in Demography. National Centre for Development Studies Australian National University Canberra. December, 1988

2 s&oaiai "Ä1

3 To my parents

4 ii D E C L A R A T I O N Except where otherwise indicated, this thesis is my own work. Canberra, December Ü Nyan Myint

5 iii A C K N O W L E D G E M E N T S I wish to express my heartfelt thanks to my supervisor, Dr. David Lucas, Director MA(D) program NCOS, for his keen interest, encouragement, supervision and invaluable criticisms while writing this thesis. I would also like to express my thanks to my advisers, Dr. Lado Ruzicka and Dr. Alan Gray, Department of Demography, for their generous help in providing me with guidance, advice and material regarding my thesis. I am also indebted and thanks to Dr. R.M. Sundrum, Department of Economics (RSPacS), Dr. J.C. Caldwell, Department of Demography, Dr. Paul Meyer, Dr. Kim Streatfield and Ms. Christine McMurray, MA(D) program NCDS, for their encouragement, help and valuable understanding to me. Similarly, Ms. Marian May, MA(D) program NCDS, for her kindness and for patience in correcting my English. I would like to express my sincere thanks to all my teachers, friendly AIDAB staff, my friends and colleagues and whoever helped me in any way. My thanks to the Burmese Government and Australian Government for the Scholarship which enabled me to study Demography at the Australian National University. I am greatly indebted to Dr. Khin Maung Nyunt, Rector, and Professor U Ko Ko, head of the Statistics Department, Institute of Economics, Rangoon, for their invaluable training, encouragement and advice in my academic career. Last but not least, I would like to express my heartfelt thanks to my parents U Su Hlaing and Daw Shu Khin, and sisters and brothers, for their constant encouragement, invaluable support, and understanding.

6 iv A B S T R A C T In Burma, levels of fertility and mortality are usually estimated directly using the data from the incomplete vital registration. These estimates, of course, are under-estimates and fertility and mortality are the areas for further research in Burma. Thus, the main purpose of this study is to estimate the levels and trends of fertility and mortality in Burma. Although the Governments of Burma have never intervened in the fertility attitude of the people, the former socialist Government limited access to any modern methods of contraception. However, this study shows that a fertility decline started in urban areas in the early 1970s. Moreover, the levels of fertility in Burma have never exceeded those in India and Bangladesh. Historically, Burma had lower fertility than India as a whole throughout the pre-war period. The higher age at marriage, the higher status of women, the higher level of female literacy and the main religion in Burma, Buddhism, which does not discourage the use of any methods of contraception, are the main causes of lower fertility in Burma. Recently, various sources, such as the United Nations, ESCAP, the World Bank and the US Bureau of the Census, have given quite different estimates of mortality, especially the level of infant mortality for Burma. Therefore, some indirect estimates of infant and child mortality, using the data on children ever born and children still living from the 1983 Census, are also made in this study. The estimates of both infant and general mortality derived in this study are much higher than both the Government's official estimates and the United Nations estimates. Moreover, this study shows that mortality has declined substantially since the mid-1950s, and the decline was much more faster in urban areas than in rural areas. However, it has apparently slowed down in recent years. The levels of mortality were much lower in Burma than in India throughout the pre-war period, and still lower than in India and Bangladesh but higher than in Thailand.

7 V Table of Contents DECLARATION ACKNOWLEDGEMENT ABSTRACT LIST OF TABLES LIST OF FIGURES Chapter Page ii iii iv vi vii Page 1. INTRODUCTION 1.1. Introduction Geographical and Demographic Background Population Censuses of Burma Vital Registration System and Statistics Evaluation of Census Data and Vital Statistics LEVELS AND TRENDS OF FERTILITY IN PRE-WAR BURMA 2.1. Introduction General Description of Pre-War Fertility Indirect Estimates of Fertility for Burma and India 2.4. Some Fertility Determinants in Burma and India LEVELS AND TRENDS OF FERTILITY IN POST-WAR BURMA 3.1. Introduction Current Levels of Total Fertility Rate Current Levels of Crude Birth Rate Current Fertility Trend and Some Fertility Determinants Government Policies on Population Growth and Fertility Comparisons with Neighbouring Countries LEVELS AND TRENDS OF MORTALITY IN PRE-WAR BURMA 4.1. Introduction General Description of Pre-War Mortality 4.3. Rural-Urban Differential in Pre-War Mortality Age and Cause-Specific Death Rates in Pre-War Period Comparison of Life Expectancy with India LEVELS AND TRENDS OF MORTALITY IN POST-WAR BURMA 5.1. Introduction Current Levels and Trends of Mortality in Urban Areas Differential Mortality Recent Levels and Trends of Infant and Child Mortality Government Plans and Policies on Health Issues Comparisons with Neighbouring Countries CONCLUSION 82 References 87 Appendices

8 vi List of Tables Table Title Page 1.1 Population, Area, Density, Percentage Urban and Intercensal Growth Rate for States and Divisions Census Date,Population, Area Covered, Comparable Population, Annual Intercensal Growth Rate and United Nations Age-Sex Accuracy Indices for the Censuses of Burma Whipple's Index, Myers' Index and United Nations Age-Sex Accuracy Index for 1973 and 1983 Censuses Male Births, Female Births, Estimated Population and Crude Birth Rates in the Pre-War Period in Burma Indirect Estimates of Fertility by Rele's Method in Burma and India, Total Fertility Rate and Crude Birth Rate by Palmore's Method in Burma and India, Child Woman Ratio, Singulate Mean Age at Marriage and Coale's Index of Proportion Married in Burma and India, Number of Towns Covered by Vital Registration and Crude Birth Rates in Urban Burma, Reported and Adjusted Values of Total Fertility Rate by States and Divisions in 1983 (per woman) P/F Ratios by Various Methods, Burma, Reported and Estimated Values of Crude Birth Rates by States and Divisions in 1983 (per 1000 mid-year population) Recent Trends and Levels of Fertility in Burma (Estimated by Palmore's Method) Some Fertility Determinants in Burma Total Fertility Rate and Crude Birth Rate for 46 Burma, Bangladesh, India and Thailand, Number of Deaths by Sex, Sex Ratio of Deaths, Census Sex Ratio and Crude Death Rate in Burma, Number of Infant Deaths and Infant Mortality Rate by Sex in Burma, (per 1000 live births) CDR and IMR by Type of Residence in Burma for Selected Years (per 1000) Age-Specific Death Rates in Pre-War Burma (per 1000) Intercensal Age Specific Death Rates in Pre-War Burma (per 1000) Deaths by Cause and Sex, and Cause-Specific Death Rate in Burma, (per 1000) Expectation of Life at Birth in Burma and India,

9 vii List of Tables (Continued) Table Title Page 5.1 Crude Death Rate, Infant Mortality Rate and Maternal Mortality Rate in Urban Burma, Indirect Estimates of Infant Mortality Rate by States and Divisions around 1982 (per 1000 live births) Crude Death Rate and Infant Mortality Rate by Sex in States and Divisions (Urban Areas), Recent Trends and Levels of Infant and Child Mortality and Life Expectancy for Burma, Urban and Rural Areas Infant Mortality Rate and Life Expectancy for Burma, Bangladesh, India and Thailand, List of Figures Figure Title Page 1.1 Burma and Its Neighbours Crude Birth and Death Rates in Burma, Infant Mortality Rate in States and Divisions,

10 1 CHAPTER ONE INTRODUCTION 1.1. Introduction It is generally accepted that there is a relationship between population growth and the economic development of a country. However, a universally accepted answer to how population growth and economic development are related has not yet been found. Some theories about development view population growth as beneficial whereas others view it as a deterrent to economic growth. Similarly, many countries believe population growth is desirable in their particular situation or stage of development, while others try very hard to reduce their population growth. Nevertheless, it is undeniable that population growth is not only causative but also an associated factor in development of a country (Ghosh, 1978: 71). Burma has been one of the countries which deems population growth to be desirable and whose Government1^ did not encourage the practice of any modern methods of family planning (United Nations Fund for Population Activities (UNFPA), 1987: 79). Yet the annual population growth rate of Burma appears never to have exceeded 2.5 per cent and the intercensal growth rate between the last two censuses was only two per cent. The growth rate seems to have been declining and it is apparently lower in Burma than in 1/ In this Chapter, 'the Government' refers to the former socialist Government, which was overthrown by the military on 18 September Since the military Government announced that it could not take care of education and social affairs, it seems the previous Government's attitude on population policies is being maintained.

11 2 Bangladesh and India, countries which have adopted family planning as an important part of their national development planning. Population growth is determined by three factors: births, deaths and migration. Since international migration in Burma is negligible, the birth and death rates determine the national population growth rate. However, the vital registration system of Burma is not good enough to provide the reliable fertility and mortality statistics from which one could determine the trend of population growth. Nevertheless, there is some evidence that the level of fertility in Burma has always been lower than that of Bangladesh and India, and it might have been even lower than that of Thailand up to the 1970s (United Nations, 1986b). Although the previous Government viewed both population growth and the level of fertility as desirable, the level of mortality as high and unacceptable (UNFPA, 1987: 79). According to the vital statistics reports, which are based only on selected urban areas, both the Crude Death Rate (CDR) and Infant Mortality Rate (IMR) have declined since the mid-1950s (Central Statistical Organization (CSO), 1985: 4). However, the level of mortality, especially infant mortality, is still very high compared with the developed countries. The estimate of the life expectancy of 60 years for both sexes is higher than that of Bangladesh and India but lower than that of Thailand (United Nations Children's Fund (UNICEF), 1988: 64). However, the vital statistics for Burma are based on the data for urban areas only and current levels of both fertility and mortality for the whole of Burma are not known with precision. This study attempts to estimate the trends and levels of fertility and mortality in Burma as a whole. The main source of data in this study is the 1983 population census. In the 1983 census, 20 per cent of the census blocks were randomly selected and the householders were asked

12 3 additional questions in a long version of the questionnaire (long form). In the long form questionnaire the birth date of last child, the number of children ever born and children still living were asked of ever-married women. However, in the census publications from the 1383 census, cross-tabulations are available only for the number of children ever born alive, the number of children still living and the number of children born alive during the last twelve months by the number of ever-married women in fiveyear age groups. These tabulations are made for each of the states and divisions, urban and rural, and for the whole country. Therefore, it is possible to study differential fertility and mortality by states and divisions, and by urban-rural residence. However, differentials by socioeconomic characteristics, such as the education level, or occupation, or the labour force participation of women or their husbands, can not be obtained from these data. Since the vital registration system does not cover the whole country (CSO, 1986: 1), indirect methods have to be adopted for estimating trends and levels of mortality and fertility, not only for the whole of country, but also for some domains such as urban-rural, and states and divisions (the main administrative regions). In addition, the levels of fertility and mortality in the pre-war period and during the last intercensal period will be estimated in this study from the census population distributions Geographical and Demographic Background Burma is situated in the Indo-China peninsula and borders with China in the north, Laos and Thailand in the East, and Bangladesh and India in the West. Burma has about 2300 kilometres of coastline in the west and south along the Bay of Bengal. The area of Burma is 676,578 square

13 4 kilometres. About 48 per cent of the area is covered by forests and more than half of the land is mountainous. Central and Lower Burma consist of valleys or flat plains with small hills, but the northern and eastern parts are hilly and mountainous. Three mountain ranges, beginning from the eastern extremity of the Himalayan mountain range, run parallel to each other from north to south. Some parts of the Western Mountain Ranges, with a height of 1000 to 3000 metres above the sea level, make a natural wall between Burma and India. Similarly, the mountains of Tibet, with an average height of 3000 metres, and the Eastern Ranges, with a height of 1000 metres, separate Burma from China, Laos and Thailand. These walls have protected Burma from huge waves of immigration along the land routes from two population giants, China and India. Since the new administration introduced in 1974, Burma has comprised of seven states and seven divisions. Under these 14 administrative domains there are 314 townships formed by 228 towns, 2190 wards and 13,756 village tracts. The major ethnic groups, in descending order by the size of population, are Burmese, Shan, Karen, Rakhine, Mon, Chin, Kachin and Kayah. The Burmese live mainly in the divisions, and the other major ethnic groups live mainly in their own states. As most parts of Burma are under deep forest and mountainous, and as Burma also has plenty of rivers and streams, the communications, not only between Burma and its neighbours but also within Burma, especially in the states are still very difficult. The population of Burma is 39.3 million in 1988 (Union of Burma, 1988: 16) and 69 per cent are Burmese. The level of urbanization is rather low: the proportion of urban population is only 23.9 per cent of the total population and only marginally greater than the 23.6 per cent in Even though the Government perceived the spatial distribution as partially appropriate (United Nations, 1987: 90), the

14 Figure Burma and Its Neighbours World1?!? f^ m ^ C y clopedia of rha Th-ird Herld (Kunan, 1982: 248) ^

15 6 population distribution is uneven: the fertile lowlands and the delta of the Irrawaddy basin are densely populated, whilst the northern and eastern parts of Burma are sparsely populated. The population density of 58 persons per square kilometre in 1988 is the third lowest in South-East Asia after Laos and Kampuchea. Population, area, population density, and percentage of urban population in 1983, and intercensal growth rates for the states and divisions are presented in Table 1.1. The three leading religions in Burma are Buddhism, Christianity and Islam. Of the total population, 89.4 per cent are Buddhists. The Christians and Moslems constitute 4.9 per cent and 3.9 per cent of the total population, respectively (Union of Burma, 1986: Part-2, 55). However Buddhists are not a majority in all ethnic groups (races). If classified by ethnicity, about 99 per cent of Burmese, Mons and Rakhines, 91 per cent of Shans and 74 per cent of Karens are Buddhists. In the other races, the Christians are a majority: 85, 63, 62 per cent of Kachins, Kayahs and Chins, respectively, are Christians. Moslems are a majority among mixed-blood people (foreign and indigenous races) and foreign races, especially among the Bangladeshis. The 1-eve 1 of literacy has increased considerably during the last intercensal period: the percentage literate in the population five years of age and over increased from 66.9 per cent in 1973 to 76.6 per cent in Most of this increase occurred among females. The percentage literates for males and females in 1973 were 76.6 and 57.4, respectively (Union of Burma, 1976: xx). In 1983, the respective percentages increased to 82.0 and 71.3 (Union of Burma, 1986: Part-1, 24). However, only 1.3 per cent of population had tertiary education, which included university undergraduate, graduate, certificate, diploma and postgraduate categories in the 1983 census. Remarkably, in 1983,

16 7 Table 1.1 Population, Area, Density, Percentage Urban and Intercensal Growth Rate for States and Divisions Region Population Area (sq.km) Density (sq.km) Percent Urban Growth ratea Burma 35,307, , Rural 26,841,621 na na Urban 8,466,292 na na States Kachin 904,794 89, Kayah 168,429 11, Karen 1,055,359 30, Chin 368,949 36, Mon 1,680,157 12, Rakhine 2,045,559 36, Shan 3,716, , Divisions Sagaing 3,862,172 94, Tenasserim 917,247 43, Pegu 3,799,791 39, Magwe. 3,243,166 44, Mandalay 4,577,762 37, Rangoon 3,965,916 10, Irrawaddy 4,994,061 35, Burmese Abroad 7,710 Sources: Union of Burma (1976: vi, 1; 1986: Part-1, 13) Note: a. The growth rate represents the intercensal growth rate between 1973 and 1983, computed exponentially, na = Not available. = Not applicable.

17 8 the number of females studying at tertiary level (61,246) exceeded that of males (60,093). For calculation of the labour force participation, Burma considers the population aged ten and over. In the 1986/87 fiscal year million or about 54 per cent of population age ten and over were in the labour force. Of the total active labour force, 62.8 per cent were employed in the agricultural sector. Trade and services employed 9.7 per cent each and processing and manufacturing employed 8.7 per cent. The other productive sectors employed 4.9 per cent and the remaining 4.2 per cent were job seekers (Union of Burma, 1987a: 20). According to the 1983 census, females constitute 36.0 per cent of the labour force; this proportion increased from 33.1 per cent in 1973 (Union of Burma, 1976: xxiv; 1986: Part-2, 79) Population Censuses of Burma In Burma, population counts were known to have taken place since 500 B.C., during the era of King Thadodipa Mahadazayarza of the Tagaung Dynasty (Nyunt, 1978: 8). In 1638 A.D., during the era of the Prome dynasty, King Thalun ( ) tried to collect information on households, land area, cropping, production and population in his kingdom. However, apart from knowing what had been done, the records of those inquests^/ (or sit-tans in Burmese), are not available (Harvey, 1947: 149). During the Konbaung dynasty ( ) the Burmese sit-tans were taken again by King Bodaw ( ) and successive rulers (Träger and Koenig, 1979: 2). These sit-tans were based on the administrative regions, but were not taken on a regular basis. They usually dealt with Harvey (1947), Hall (1950), and Träger and Koenig (1979) use the word inquest for the sit-tan in their articles, and the same term is also used here.

18 9 information on households, manpower, agricultural output and revenue (Union of Burma, 1983: 1). Using these sit-tans, Burney (1842: 336) estimated the population of the Burmese empire in 1783 and 1828 as 4,209,240 and 4,230,558, respectively. He estimated the population indirectly from the estimated number of houses (which was ten per cent more than recorded), by allowing, on average, seven persons to a house. Although his estimation method was very crude, Maung (1986: 5) remarked that those first estimates of Burmese population were plausible. The first modern population census was carried out in 1872 under the British administration, as a part of the Indian census. Thereafter censuses were taken every ten years starting from The 1872 and 1881 censuses covered only Lower Burma^. After the annexation of Upper Burma in the third Anglo-Burmese War in 1885, the 1891 census and subsequent censuses were taken for the whole of country. Regular decennial census taking was carried on up to The reports of these colonial censuses involve many interesting interpretations of the data. For example, in the section on the "Causes of Low Density", the 1911 Census Report (Webb, 1912: 11) explained it as:...the isolating effect produced by enveloping mountain ranges. Even now, between the teeming multitudes of the Indian and Chinese plains and the comparatively empty valley of the Irrawaddy, there intervenes a dreaded unknown hill country inhabited by untamed hill tribes, operating as a barrier to any but the most fragmentary degree of intercourse. These Census Reports are not only a systematic recording of population, but are also the historical records of Burma. They covered a wide range of analyses and historical events. The administrative reports (in the administrative volumes) provide the best information on "error and internal anomaly" 3/ Lower Burma consists of the Irrawaddy, Pegu, Rangoon and Tenasserim divisions, and the Mon and Rakhine states.

19 10 (Martin, 1981: 62). The 1921 Census Report, for example, recorded the great influenza epidemic during and analyzed its impact as follows: Influenza thus accounts for a reduction by about 2.85 per cent of the rate of increase for the decade in the registration area (Grantham, 1923: 34). The 1881 and later Census Reports include also the actuarial reports. These reports were prepared using census returns and vital statistics. They include the expectation of life at selected ages, recorded birth and death rates. leading causes of death, and From the 1921 census, special attention was paid to industrial and economic information and statistics. In the 1931 census, a special fertility inquiry was attempted to estimate the rate of reproduction in various social classes (Natarajan, 1972: 518). The 1941 census was the first in which Burma was taken separately from India. In that census, international migration was fully examined. But, except for the provisional total population by district, all the records from the 1941 census were lost during World War II (Lin, 1976: 40). In these colonial censuses, the area covered varied (Nyunt, 1978: 9) and it is difficult to study population growth and changes in population composition. The methodology of taking these censuses was explained by Sundrum (1957: 2) as a combination of three procedures: (a) in the fully administered areas, a synchronous de facto method or enumeration of persons who were actually resident in those places as of midnight on the census date; (b) in the loosely administered and sparsely populated areas, the de jure method or the enumeration of all people normally resident in those areas; and (c) for the remaining parts of the country, estimates were made by the administrative officers.

20 11 The most interesting point concerning the colonial period population is the relationship between migration, sex ratio and date of the census. Except for the first census which was taken during August, all censuses of the colonial period were taken during February or March, which was just after the harvest. Although the sex ratio for the indigenous population was less than 100, for the country as a whole it was more than 100 (Sundrum, 1957: 23); that is, the male population exceeded the female population throughout the colonial period. This excess of males was mainly because of immigration of Indians and Chinese. The majority of migrants, about eighty per cent of the foreign population in Burma, were Indians. Most of them were seasonal workers in the harvesting, milling, and shipping of rice: they were predominantly males and the number of immigrants was greatest at the time of the census. The sex ratio of the Indian population in Burma was more than 300 in every census. However, the Indian immigration to Burma was largely temporary; the ratio of emigrants to immigrants was about 80 per cent every year (Davis, 1951: 100). The second point of interest is the population distribution. The delta area (Irrawaddy and Rangoon divisions) which was a newly settled area at that time, had developed quickly and became the most densely populated area in Burma. Three supporting factors for the quick development and relatively high density in the delta area were: firstly, a large amount of internal migration from the other parts of Burma to obtain fertile land in the reclamation of the delta after the opening of the Suez Canal in 1869 (Furnivall, 1957: 47); secondly, seasonal migration from central Burma to the delta especially in the cropping season; and finally, the concentration of immigration from India and China in the delta area (Sundrum, 1957: 33).

21 12 The third point to mention here is the link between international migration and urbanization in Burma. International migration was not clearly identified in any of the colonial period censuses but total population was expressed in terms of indigenous and foreign population. Urbanization during the colonial period was heavily dependent on the migration of Indians and Chinese. During the period the urban population increased by about half a million, and the percentage of urban population increased from 9.5 to 10.4 per cent. More than 60 per cent of the increase in urban population was contributed by foreigners. In 1901, about 31 per cent of the urban population was foreign born and it grew very rapidly to a maximum of 41.5 per cent in 1931 (Sundrum, 1957: 22). The development of some of the towns in Lower Burma depended mainly on international migration, especially Indian migration. Rangoon, for example, developed very rapidly during the colonial period. When Lower Burma was annexed by the British in 1852, only about 30,000 people were living in Rangoon. By 1901, it had become a city with a population of one-quarter of a million. However, of Rangoon's population, the indigenous population was only about 33 per cent with the Indians constituting 48 per cent. Thereafter, the foreign population in Rangoon was more than twice the size of the indigenous population and the Indian population alone was more than one and a half times the indigenous population (Chakravarti, 1971: 9-19). Burma gained her Independence on 4 January Conditions were unsettled throughout the country soon after Independence and the regular census, due in 1951, could not be taken. In 1953 there was an attempt to take the first post-independence census by stages, stretching over the period The urban census of 1953 was successfully conducted and covered about 99 per cent of the total urban area of that time (Union of Burma, 1975: 11). The rural

22 13 census of 1954 covered only 15 per cent of the rural population was a sample survey because the villages were chosen from the point of view of convenience and security of the enumerators. It was intended to continue the census for the rest of the rural areas in subsequent years, but the attempt failed. To close the gap of about 30 years without any country-wide census, a census was conducted in April 1973 on a de-jure basis as the first nation-wide census after Independence. One purpose of the census was to obtain reliable population data, but the primary purpose was to prepare the Electoral Rolls for the Constitutional Referendum in 1973 and the Constituent Assembly in 1974 (Nyunt, 1978: 13). The census covered about 85.1 per cent of total area and 97.1 per cent of the total population (Lin, 1976: 40). It was the first census in Burma which gave population by single years of age. However, neither fertility nor mortality related questions were asked in this census. The census procedure and questionnaire design, enumeration, organization and the tabulation are fully discussed by Lin (1976) and Nyunt (1978). Exactly ten years after the 1973 census, Burma conducted another census. This census covered 96.6 per cent of total population (Union of Burma, 1986: Part-1, 11). In this census two types of questionnaire were used, a short form and a long form. The short form, which was asked of 80 per cent of the total population, includes only seven basic questions: name, relationship to the head of the household, sex, age, marital status, race and religion. The long form was administered to the remaining 20 per cent, which was selected randomly. The long form questionnaire included, in addition to the seven short form questions, an additional eleven questions concerning demographic, socio-economic and fertility characteristics.

23 14 The additional questions in the long form are on school attendance, highest standard passed, literacy, occupation, industry, employment status, reason for not working, whether working during the last twelve months, children ever born alive, children still living, and date of birth of the last child. Based on the information from the long form questionnaires, 27 tables including seven fertility and mortality related tables were presented in the census publication. All data in these tabulations are weighted to match the total population count based on ruralurban areas, age and sex (Union of Burma, 1986: Part-1, 5). As mentioned in the introductory section, the cross-tabulations concerning fertility and mortality are very limited: they present only the number of children ever born alive, the number of children still living and the number of children born alive during the last twelve months by the number of ever-married women in five-year age groups for each state and division, and for rural and urban residence. Although the number of ever married women by the number of children born alive is cross-tabulated by literacy, highest standard passed, and labour force participation, none of these tabulations is by the age of women nor with respect to children still living. Therefore, the study of mortality can not proceed to include the other differentials by, for instance, education level, occupation and labour force participation of women or their husbands. Nevertheless, the 20 per cent sample of the 1983 census gave reasonable estimates of the levels and trend of fertility in Burma. Since the decline in fertility between 1973 and 1983 was not recognized, fertility after 1973 had been over-estimated until the analysis of the 1983 census deflated previous estimates. Because of the over-estimation of fertility, the population growth rate and the total population had also been over-estimated. The population growth rate was accepted to be as high as 2.3 per cent

24 15 during the intercensal period, but it was corrected to only 2.0 per cent after the 1983 census. Consequently, the intercensal population estimates were wrong, and the socioeconomic indicators, associated with population size in various reports, were wrong. Similarly, the United Nations population projections for the year 2025 had to be reduced: from 82 million in the 1982 assessment to 66 million in the 1984 assessment (United Nations, 1985: 380; 1986b: 266). Both the 1973 and 1983 censuses adopted the de jure method. The censuses were taken during 1-5 April and the census date was set to be midnight of 31 March of the respective years. Enumeration teams were sent prior to the census date to hill tracts and remote areas where communication was difficult. If necessary, extension of the enumeration period beyond the scheduled date was allowed. The household population was enumerated on the first four days and the moving population, such as the homeless and the boat dwellers, were enumerated on the last day. However, some parts of the country were omitted from enumeration for security reasons. The estimated population in those areas in the 1973 and 1983 censuses was 836, 713 (2.9 per cent) and 1,183,005 (3.4 per cent) respectively (Union of Burma, 1976: viii; 1986: Part-1, 12) Vital Registration System and Statistics In Burma, a vital registration system to collect birth and death statistics was first introduced in 1864 in a part of Lower Burma (Maung, 1986: 23). It was introduced into the towns of Upper Burma in 1906 and the villages of Upper Burma in After that it covered nearly 80 per cent of the total population; in 1931 about 82.5 per cent of the population was covered (United Nations, 1959: 47). In the pre-war period (period before World War II) vital statistics were collected by Municipal Health Offices in

25 16 urban areas and by the village headman in the rural areas. The reports were published by the Department of Public Health (Sundrum, 1957: 7). Vital registration was completely stopped during the War and revived in the post-war period. A new system of vital registration, starting with Rangoon City and 15 other towns, was introduced on 8 February It was gradually extended to the other towns and by the end of 1981, 243 out of 288 towns were covered (CSO, 1986: 1). However, the vital registration system for the rural areas covers only 25 out of 314 townships. According to the Vital Statistics Report (CSO, 1985: 1), in 1980 the vital registration system covered 88 per cent of total urban population and only 6.1 per cent of total rural population. Thus, the vital registration covered not more than 27 per cent of the total population. Nowadays, the vital statistics are collected under the authority of the Directorate of Health Services (formerly the Department of Public Health), and compiled and published jointly by the Central Statistical Organization (CSO). However, the vital registration system covers births (live births and still births) and deaths only. Moreover, the latest available Vital Statistics Report is for the year 1981, more than six years out of date (CSO, 1986) Evaluation of Census Data and Vital Statistics Kingsley Davis (1951: 4) remarked on the reliability and usefulness of the Indian censuses, which covered Burma until 1931, as follows: Compared with the world as a whole, India's past population statistics are above the average... are remarkable not only for the information they reveal but for the special obstacles they have had to overcome.

26 17 Also the administrative efficiency of these censuses could be seen in the speed of the report publishing. The reports were published not long after the date of census without electronic machines. The superintendent province), mentioned the age-reporting as: of the Census of India (Burma... probably a closer approximation to the actual facts than those obtained in any other province of the Empire; in fact, in the matter of accuracy, not far behind those of European countries (Lowis, 1902: 67). Similarly, in the section of "Accuracy of the age statistics" of the 1921 Census Report, Grantham (1923: 122) concluded that though the age statements were defective, they were approximately correct if combined for five-year age groups. However, the census coverage varied from year to year and none of the colonial censuses gave the population by single years of age. To evaluate the colonial period censuses by modern methods, the United Nations Age-Sex Accuracy Index is computed for the years 1891 to The census date, population, area covered, the estimated population in the comparable area of the 1973 census and the annual intercensal growth rate and the United Nations age- sex accuracy index are presented in Table 1.2. From Table 1.2, it can be seen that the United Nations age-sex accuracy index for the 1931 census is only 25. Therefore, the age reporting was reasonably accurate. However, the age reporting of the other censuses was rather rough, since the indices exceed 40. The rise in the index for the 1921 census could be partly because of the influenza epidemic of , when many children and old people died. Compared with the Indian censuses, all Burmese censuses are better in terms of age reporting. The same indices for the whole of India are around 80 except for the year 1931 (only 26.2), which is unexplainably low.

27 18 Table 1.2 Census Date, Population, Area Covered, Comparable Population, Annual Intercensal Growth Rate and United Nations Age-Sex Accuracy Indices for the Censuses of Burma Census Date Population (million) Area sq. km Comparable Population Growth Ratea UN Index Aug b na na Feb b na na Feb c na Mar na Mar Mar Mar Feb na 1953d e na 1963d e na Mar Mar Sources: Union of Burma (1976: viii; 1986: Part-1, 1). Notes: The comparable populations (population living in the 1973 census area) are from the Demographic Report (Union of Burma, 1975: 16). a. Annual Exponential Growth Rate. b. Census covers only Lower Burma. c. Census covers Lower Burma and only some part of Upper Burma. d. Estimates from the 1963 Demographic Year Book (United Nations, 1964: 152). e. Assumed to be the same as 1973 area, na = Not available.

28 19 Apart from variation in the census coverage and the usual content errors, such as age mis-reporting, one of the most important sources of content error in the pre-war censuses was the inappropriateness of census definitions (Sundrum, 1957: 5). Most of the census definitions were of greater relevance to the Indian sub-continent than to Burma, because the religious, social and economic conditions of Burma were quite different from those of the rest of the Indian sub-continent. Moreover, some definitions were very crude; for example, in the 1921 and 1931 censuses, all persons who belonged to one of the races of India or to an unspecified race of India, or who spoke one of the Indian dialects, were classified as Indians. Thus, the Rakhine Moslems who speak Chittagaung dialect, and mixed-blood persons who could not specify their exact race, were classified as Indians although for all purposes they were Burmese (Chakravarti, 1971: 13). Age reporting in the post-independence censuses improved. However, age reporting in the 1983 census was not as good as in the 1973 census. Whipple's index, Myers' index and the United Nations age-sex accuracy index for urban and rural areas and for the whole country for both censuses are presented in Table 1.3. The possible range of Whipple's Index is from 100 to 500 depending on the final digits of the reported ages are whether evenly distributed or only 5 and 0. The Whipple Indices given in Table 1.3 show that the age reporting in the urban areas in the 1973 census was highly accurate (less than 105), but the 1983 census data were only approximate since the indices lie between 110 and 125. The Whipple Indices for the rural areas in both censuses, which lie between 125 and 175, also point out that the age data were rough (United Nations, 1962: 17). However, the United Nations Age-Sex Accuracy Indices show that the age reporting

29 20 Table 1.3 Whipple's Index, Myers' Index and United Nations Age-Sex Accuracy Index for 1973 and 1983 Censuses Area 1973 Males Females 1983 Males Females Whipple's Index Burma Urban Rural Mvers' Index Burma Urban Rural UN Index Burma Urban Rural Sources: Union of Burma (1976: 24; 1986: Part-1, 19; 1986: Part-2, 23). in all areas in both censuses was accurate since all indices are less than 20 (Shryock and Siegel, 1976: 126). The Myers Index is an estimate of the minimum proportion of the persons in the population who reported an incorrect final digit. The possible range of Myers' Index is zero, if no heaping exists, and 90, if all ages were reported at a single digit (Shryock and Siegel, 1976: 118). As shown in Table 1.3, the age reporting in urban areas in

30 21 the 1983 census was fairly reliable since only six per cent of the population gave a wrong final digit. However, about 18 per cent of the rural population reported their ages with an incorrect final digit. Generally, the age reporting was fairly accurate, especially in the urban areas, in both censuses. To assess the reliability of the 1983 census, a Post Enumeration Check (PEC) was attempted on 1 April, 1984, exactly one year after the census. This PEC was the first in Burmese census history and covered 87.7 per cent of the census area. The PEC showed that the census coverage was 99.1 per cent, and that 95.6 per cent of the matched persons (census and PEC) had the same age (Union of Burma, 1986: Part-1, 45). Moreover, it revealed that the underenumeration of children aged 0 to 4 years was about three per cent (Union of Burma, 1987b: 4). The birth and death statistics obtained from vital registration are, as pointed out earlier, not reliable in Burma. They are defective mainly in two ways: firstly, the vital registration system does not cover the whole country, and secondly, even in the area under registration, the reporting of births and deaths is incomplete. In the pre-war period, the vital registration system covered about 82 per cent of the population. But nowadays it covers less than 30 per cent of the total population. Moreover, the completeness of death registration in the urban areas, where the vital registration system has been established, was only about 70 per cent during the period (Hpu, 1984: 126). The vital registration system is just being introduced in some rural areas. Thus, fertility and mortality statistics have to be estimated from the census data, which are reliable.

31 22 CHAPTER TWO LEVELS AND TRENDS OF FERTILITY IN PRE-WAR BURMA 2.1. Introduction Fertility is defined as "actual reproductive performance" or simply "the number of live births" and more easily and usually measured for women (Lucas, 1980: 64). Commonly used measures of current fertility are the Crude Birth Rate (CBR), Age Specific Fertility Rate (ASFR), Total Fertility Rate (TFR), Gross Reproduction Rate (GRR), Net Reproduction Rate (NRR), Intrinsic Birth Rate (IBR) and General Fertility Rate (GFR). The basic data required to compute these measures are the total number of live births or number of live births classified by the age of mother. Thus, the quality of these fertility measures is highly dependent on the quality of birth data collected. On the other hand, the children ever born (CEB) or average parity is the most commonly used measure of cohort fertility. It is usually obtained from cross-sectional studies such as census and surveys, and independently of the registration system. In Burma the only available fertility statistic during the pre-war period was the CBR. However, as mentioned in Chapter One, although the vital registration system had covered more than 80 per cent of the total population in those days, it failed to give a number of live births for estimating the CBR. Moreover, the birth registration form, which was used during the pre-war period, included only limited information such as the date and place of birth, the names of the parents and the occupation of the father (Sundrum, 1957: 7). It did not include the most important information, such as the age of the women at the time of

32 23 giving birth and other information on her fertility. Hence, neither the age pattern of fertility nor the TFR could be estimated. Therefore, in this chapter the levels and trend of fertility in Burma during the pre-war period were studied using not only the CBR given by the vital registration, but also indirect estimates of CBR and TFR. The study of the levels of fertility during the pre-war period by indirect methods could not be done except for the country as a whole, because the administrative divisions during the pre-war period were changing and the population distribution by rural-urban residence was not available in most of the censuses. However, to compare the levels of fertility with India and to examine the implications of the differential fertility, the Singulate Mean Age at Marriage (SMAM) and Coale's index of proportion married (Coale, 1967: 205-9) were computed for Burma and India General Description of Pre-War Fertility During the pre-war period, vital statistics or the statistics on births and deaths were collected under the authority of the Department of Public Health. The Department computed some vital statistics such as the Crude Birth Rate, Crude Death Rate, Age-Sex Specific Death Rate and Cause Specific Death Rate. Starting from the 1881 Census Report, the actuarial reports were prepared and the vital statistics were given. Sundrum (1957, 16) compiled the number of male births, female births and the Crude Birth Rate from the census reports for selected years. However, in computing the Crude Birth and Death Rates for the intercensal years, the estimated populations were taken as the population of the registration area at the last census. There was no attempt to estimate the population

33 24 actually at risk in the intercensal years (Sundrum, 1957: 8). Consequently, the CBR for the intercensal years were unreasonably high. Thus, in Table 2.1 an exponential interpolation is attempted to get the reasonable population estimates for the years 1915, 1925 and Then the CBRs were re-computed. However, the area covered by the vital registration system before 1907 was expanding rapidly and different from one year to another. Therefore intercensal years prior to 1910 were omitted from the adjustment procedure. The male births, female births, estimated population and the Crude Birth Rate for selected years are presented in Table 2.1. From Table 2.1, it can be seen that the number of male births exceeded the number of female births with the sex-ratio at birth varying from in 1931 to in However, for the whole period from 1881 to 1935, the sex-ratio at birth was 105.3, that is, approximately 105 male births for every 100 female births. Since the sex ratio at birth is about 105 males to 100 females (Pollard et. al., 1981: 17), the five per cent excess male births during the pre-war period could be said as normal. In Table 2.1, the extremely low CBR up to 1901 showed the under registration of births with increase in CBR throughout until 1905 showing improved registration. However, according to the 1911 Census Report (Webb, 1912: 43), over 20 per cent of births remained unrecorded in the city of Mandalay although a high degree of accuracy was expected there. Thus, no reliance can be put on the CBR computed from the vital statistics and alternative estimates are needed to study the levels and trends of pre-war fertility. One of the main reasons for under reporting of births was high infant mortality during the pre-war period. When a baby died soon after its birth, it was rarely

34 25 Table 2.1 Male Births, Female Births, Estimated Population and Crude Birth Rates in the Pre-War Period in Burma Year Number of Births Male Female Sex Ratio (birth) Estimated Population CBR (per 1000) Lower Burma 1872 na na na 2,665, ,012 35, ,692, ,059 45, ,653, ,124 46, ,596, ,850 63, ,516, ,099 85, ,546, ,840 92, ,568, All Burma , , , 878, , , ,226,394a 33.9b , , ,771, , , ,285,095a 24.3b , , ,102, , , , 806,774a 29.0^ 1941 na na na 16,824, Sources: Sundrum (1957: 16-18, 45). Notes: a. Population of the year is an exponential interpolation of two adjacent census populations. b. The CBR given in the source for the years 1915, 1925 and 1935 is 35.1, 25.5 and 33.0 respectively. c. From 1981 Vital Statistics Report (CSO, 1986: 4). na = Not available.

35 26 reported. Moreover, some parents did not want their babies to be vaccinated and they did not report the birth since the health officers vaccinated the children whose births had been reported (Sundrum, 1957: 8) Indirect Estimates of Fertility for Burma and India As mentioned in the above sections, the birth registration was inadequate in terms of both information available and accuracy of data. Therefore, alternative estimates are necessary to study the levels and trend of fertility during the pre-war period. In this study, two indirect techniques, namely Rele's Method (Rele, 1967; 1987; 1988) and the Palmore Regression Method (Bogue and Palmore, 1964; Palmore, 1978), were used to estimate the levels and trends of CBR and TFR during the pre-war period. Since Burma was a province of India from 1885 to 1936, the levels of fertility in India during the same period are also estimated. The main reason for chosing Rele's and Palmore's methods is that these methods required only the data available in the pre-war Census Reports and are not restricted by the stable or quasi-stable assumptions. (a) Rele's Method The Rele method is an extension of the stable population concept; however, it is not necessary to assume population to be stable, quasi-stable or closed to migration (Rele, 1987: 514). The main advantage of the method is that it needs only Child Woman Ratios (CWR) and an approximate estimation of life expectancy (e0), both of which are usually available from published sources. Moreover, the method is not very sensitive to the estimate of life expectancy: a change of one year in e0 has an effect of less than 0.8 per cent on the estimated value of GRR, TFR or CBR