Senegal - Population sector report (Vol. 2 of 2) : Technical note

Report No. 6510-SE Senegal Population Sector Report (In Two Volumes) Volume II: Technical Note June 24,1987 Population, Health and Nutrition Department FOR OFFICIAL USE ONLY H~~~~~~~~~~~~~~~~~~~~~~~~~ Document of the World Bank This report has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  FOR OMCIAL USE ONLY SENEGAL POPULATION SECTOR REPORT VOLUME 1I: TECHNICAL NOTE Table of Contents PMae No. I. SOURCRS AND QUALITY OF DATA . . . . . . . . . . . . . . . . 1 II. MORTALITY . . . . . . . . . . .. . . . . . . . . . . . 2 A. Introduction.. . . . . . . ..2 B. Estimation of Base Mortality Levels for Projection . . . 3 1. Estimation of Mortality from Child Survival Data ............. . . . . .... . 3 2. Estimation of Child Mortality from the SPS Maternity Histories . . . . . .11 III. FERTILITY .. . . . . . . . . . . . . . . . . . . . . . . 14 A. Introduction ...14 D. Estimation of Base Fertility Levels for Projection . . . 15 This report is based on the findings of a World Bank mission which visited Senegal in May-June 1985 in conjunction with a UNFPA mission, which colaborated closely with the World Bank and which has prepared a separate report. World Bank mission members included Mr. A. Williams (Mission Leader), Ms. A. Hill (Demographer) and Ms. D. Benjamin (Information, Education and Communications Consultant). UNFPA mission members included Dr. Sabwa a Matanda (Public Health Specialist), Mr. M. Mazouz (Demographic Research and Training) and Ms. S. Crapuchet (Women's Issues). A second World Bank mission, comprised of Mr. A. Williams (Mission Leader), Mr. M. Azefor (Demographer) and Ms. D. Vaillancourt (Operations Analyst), visited Senegal in March 1986 to present preliminary conclusions of the sector work and to refine its recommendations. The main sector report has been prepared by Ms. D. Vaillancourt; the technical note in Volume II has been prepared by Ms. A. Hill. This document has a restricted distribution and may be used by recipients only in the perfomance of their oMcal duties Its contents may not otherwise be disclosed without World Bank authorization.  - ii - SENEGAL POPULATION SECTOR REPORT VOLUME Il: TECHNICAL NOTE Glossary of Acronyms ASAFED : African Association for Development Education ASBBF Senegalese Association for Family Well-Being BNR : National Census Bureau BOPP Protestant Social Action Organization OONAPOP : National Population Commission CBR : Crude Birth Rate CDR Crude Death Rate CEDPA Center for the Development of Population Activities CIRST Interministerial Council on Scientific and Technical Research B$C : National Economic and Social Council DID Division of Surveys and Demography DRH: Division of Human Resources FAFS Federation of Women's Associations of Senegal FP Family Planning ENC Information, Education and Communications Activities iFAN Fundamental Institute for Black Africa IPPF International Planned Parenthood Federation IUD Intra-Uterine Device WCH Maternal and Child Health MOPC Ministry of Planning and Cooperation MOSD : Ministry of Social Development MODF : Ministry of Economy and Finance MfPH : Ministry of Public Health mGO Non-Governmental Organization GRANA : Research Group on African Food and Nutrition ORTS National Radio and Television Network of Senegal SSD Secretariat of State for Decentralization SFS Senegalese Fertility Survey UNFPA United Nations Fund for Population Activities USAID United States Agency for International Development TFR : Total Fertility Rate WFS : World Fertility Survey  SENEGAL POPULATION SECTOR REPORT VOLUME II: TECHNICAL NOTE DERIVATION OF MORTALITY AND FERTILITY ESTIMATES FOR PROJECTION PURPOSES 1. This note details the derivation of estimates of base levels and patterns of mortality and fertility in Senegal, for use as a basis for population projections. It incorporates the results of a detailed demographic analysis carried out for the Bank by Douglas Ewbank at the University of Pennsylvania. I. SOURCES AND QUALITY OF DATA 2. Senegal has so far held only one population census, in 1976, although a second is currently planned for 1988. In addition, three national demographic surveys were carried out in 1960-61, 1970-71 and 1978 (the latter being one of the World Fertility Survey series), as well as a 1970 Labour Force Survey which collected data on migration. Several small- scale surveys, chiefly focussed on mortality, have also been held over the past 30 years in various parts of the country, including the Fleuve, Sine- Saloum and Oriental Regions. Another has recently been undertaken in Thies Region, and a series of regional surveys is planned for the future. Vital registration does not exist for practical purposes, except in Dakar, where completeness of registration is aufficiently high to permit some use of the data. A small UN project is planned to improve vital registration, but is unlikely to make any impact at a national level in the foreseeable future. 3. Not all of this multitude of demographic data collected has been processed or published. The results of the 1960-61 and 1970-71 surveys have been only partially analysed and published. Data from the 1976 census and the 1978 Senegal Fertility Survey (SFS) have been published, but analysis of the former is not yet completed. Most of the small-scale survey data, however, have appeared in priit in some form or other. 4. Data quality is variable. The 1976 census data appear in general to be of reasonable quality, although the Post Enumeration Survey, designed to check on completeness of enumeration, unfortunately did not yield any useful results. The quality of the 1960-61 and 1970-71 surveys is uncertain, due to incomplete processing and publication, and appears suspect in 1960-61 at least. The 1978 SFS yielded a mass of detailed data; much of it appears to be of good quality, but there are some serious inconsistencies in the results, both internally and in comparison with other data sources, that cast doubt on the reliability of at least some of the data. The small- scale surveys in general appear to have produced good and reliable data, but their usefulness is limited because they are not representative and their samples are small. Evaluation of vital registration data is always difficult and uncertain, but the Dakar data are generally considered to be usable with caution. These inadequacies in the available data clearly point to an urgent need for further census and national survey data.  II. MORTALITY A. Introduction 5. Levels and trends of mortality in.Senegal are uncertain. Although a plethora of data are available, there are serious discrepancies in the results. Generally speaking, small scale surveys held between 1957 and 1980, mostly in rural areas, have reported extremely high mortality levels, with infant mortality iates (IMR) ranging from 160 to 200, with 36 to 50 percent of children dying during their first five years of life, and life expectancies well below 40 years. Declines in mortality have bee extremely minor or nonexistent up to the mid 19709, but then a marked fall has been noted in some cases. In one longitudinal study in Sine-Saloum, for example, the IMR fell from around 200 to 150 during the late 1970s, after little or no change between 1965 and 1975. The credibility of these reported mortality levels and trends is reinforced by very similar levels and trends in Gambia, collected both in the 1973 census and in the well-known, small- scale, longitudinal Keneba study running from 1947 up to the present day. 6. Urban levels of mortality, of course, were probably always much lower. The recorded IMR from registration data in Dakar, for example, Was 108 in 1955, 80-90 in 1970, and around 60 in recent years, although probably true levels were somewhat higher at every period because of incomplete reporting. Combinations of these small-scale rural and urban estimates could be expected to yield a national mortality level that is still high by world standards and by comparison with most of contemporary eastern and southern Africa, though not unprecendented for contemporary West Africa: the IMR might be put in the neighborhood of 150 by the late 19709, with about 36 percent of children failing to survive their first five years of life and life expectancy around 40 years. 7. National survey data, however, show significantly lower levels of mortality and conflicting trends. As detailed in later paragraphs, data from the 1960-61 survey on child survival reported by mothers indicate that about 40 percent of children were dying before age 5 in the early 19408, but that this proportion had fallen to around 33 percent by the late 1950s. Life tables derived from deaths reported to have occurred during the year prior to the 1970-71 survey give an DMR of 109 and a life expectancy of 44 years at that date. In both these cases, the quality of data is uncertain or suspect. 8. More seriously, however, data from the 1978 SFS survey are also out of line with the small scale survey results. As detailed later, data on child survival reported by mothers in the initial household survey indicate that the proportion of children dying before age 5 fell from about 30 percent to about 27 percent during the 1960s and then remained at around 28 percent up to around 1975. The combination of 1960 and 1978 child survival data thus show a steady and substantial overall decline in child mortality from the 1940s to around 1970, with wastage in the first five years of life dropping from 40 to under 30 percent by 1970, followed by a plateau in the early 1970s. Direct reports of child deaths from the SFS, however, given in the course of detailed maternity histories elicited during subsequent  -3 - individual interviews with women, (smoothed by moving 5 year averages) show the proportion dead before 5 years staying rather constant between 1950 and 1970, at a level of 28-29 percent, then dropping steeply to around 20 percent by 1978. Thus, neither the trends nor the levels agree (except for a short period in the late 19609) between the two sources of child mortality data in the 1978 SFS, and both indicate levels of mortality lower than do the small-scale surveys. 9. It is impossible to choose with any certainty among these various possibi.lities. However, it seems unwise to reject wholescale such a detailed, representative, and carefully collected body of data as the 1978 SFS in favour of scattered and small-scale surveys. Of the two sources of data within SFS, the household survey child survival data seem preferable since the mortality levels indicated are the most consistent with other information (including the 1960-61 child survival data). They also provide a coherent trend of decline in childhood mortality over the post-war period; such a trend is likely at least for the more urbanized parts of the country, even if not for the rural areas where most of the small-scale surveys were held. B. Estimation of Base Mortality Levels for Projection 10. A life table for Senegal around 1970 was constructed by Ewbank, using national child survival data and small-scale survey data as detailed below. 1. Estimation of Mortality From Child Survival Data 1. The estimation of child mortality from child survival data from the 1960 and 1978 surveys is based on a computer program of a various set of Brass-type multipliers produced by Brass, Sullivan, and Trussell. There are several advantages to returning to the original equations rather than depending on any of the sets of estimation equations. First, it is possible to base the estimates on any model life table, not just the Coale and Demeny or the Brass tables for which published multipliers are available. This is a big advantage for Senegal since its age pattern of mortality is very different from all of the published standards. 12. Tn order to develop a standard life table for use in the estimation, we have combined the following life tables with the given weights: 1) data for the Fula Bande and the Niokholonoko in the Bandafassi area of eastern Senegal (from Pison and Langaney, "The Level and Age Pattern of Mortality in Bandafassi (Eastern Senegal): Results from a Small-Scale and Intensive Multi-round Survey," African Demography Working Paper No. 11, University of Pennsylvania, June 1984) - weight of 0.15. 2) data for the Ngayokheme area of Sine Saloam (from Garenne, "The Age Pattern of Infant and Child Mortality in Ngayokheme  -4- (Rural West Africa)," African Demography Working Paper No. 9, October 1981, University of Pennsylvania) - weight of 0.25. 3) an estimated life table for Dakar based on Brass' African standard transformed with legit to give an IMR cf 60 and extra mortality added at ages 1-4 to match the reported rate - weight of 0.25 to represent urban areas, and 4) an estimate of mortality for other areas designed to match the common pattern of excess mortality at ages 1-4 with the overall level of mortality determined so that the resultiug national life table is consistent with that from preliminary child survival estimates - weight of 0.35. 13. The life table derived in this way is presented in Table I. It is very similar to the life table based on deaths in the last year reported in the 1970 demographic survey. Because all of the available data for various parts of the country and for various time periods show very similar age patterns of mortality, the estimated age pattern of mortality is not sensitive to the exact weights given to the various data sets. 14. The second advantage of returning to the original integration equations is that the fertility curve used in the integrations can be based on information other than the PARI and PAR2 used by most of the existing sets of equations for estimating the multipliers. Since the integration program uses the Coale-Trussell formulation for the fertility schedule, any information (including the PARI and PAW2 ratios, reportd of recent births or registered births) can be used to estimate the parameters of the fertility schedule. This proved to be very useful since the parity ratios from both the 1960 and 1978 surveys lead to unreasonable values of ao and k, the basic parameters of Coale and Trussell's fertility curve. However, the age- specific fertility curve derived by averaging the values of ao and k from the two surveys (roughly 11.5 and 0.5) is quite similar to the pattern given by the reports of births in the previous year from both the 1960 and the 1978 surveys. 15. The third advantage is that the integration formulas can be modified to incorporate a standard mortality schedule that is specific for the age of the mother at the time of the births. For this purpose we have used a schedule which when linked with the assumed age-specific fertility schedule for 1978 provides the following schedule of relative infant mortality rates: 15-19 20-24 26-29 30-34 35-39 40-44 45-49 1.18 0.92 0.88 0.96 1.05 1.13 1.22 These estimates are consistent with data on IMR by parity from Cantrelle and Leridon for Niakhar: 1 2-3 4-6 7+ 1.16 0.89 0.90 1.16  - 5 - Table I PREPARATION OF STANDARD LIFE TABLE FOR SENEGAL. 9/18/84 DAKAR PISON Ngayokheme 1972-81Est. Rest Combined Q(x)+ QK lx Q(x) Qlx) Qx Is Weigh .25 .15 '63-'81 .25 .35 AGE ------