Luy_englisch.indd A Classifi cation of the Nature of Mortality Data Underlying the Estimates for the 2004 and 2006 United Nations’ World Population Prospects Marc Luy Abstract: The estimation of mortality conditions and trends is a sophisticated task for most populations in the world, above all for those of developing countries. After two decades of intensive discussion and derivation of specifi c estimation tools for these populations, the use of indirect estimation techniques seems largely forgot- ten among those who are not forced to apply them. However, for the majority of developing countries these methods are still the main and often the only available estimation tool. In order to systematise the available data and applied estimation techniques, we developed a fi ve-scale classifi cation of the nature of mortality data and assigned all countries with more than 100,000 inhabitants to the corresponding groups. The classifi cation is based on three sources of information regarding the nature of mortality data, the analytical reports of the 2004 and 2006 revisions of the United Nations’ World Population Prospects and the methods and data descriptions of the 2006 Global Burden of Disease Study. Although the information provided by our classifi cation is purely descriptive rather than giving a detailed overview of the specifi c methods and approaches, the contents of this paper should be of interest to politicians and scientists using the United Nations’ World Population Prospects as well as to scholars who teach and learn about indirect demographic estimation techniques. Keywords: mortality · nature of mortality data · life expectancy · indirect estimation techniques · model life tables · world population prospects · orphanhood method · growth balance method 1 Introduction The biennially updated “World Population Prospects” (WPP) of the United Nations form probably the most extensively used source of worldwide demographic data. This is well justifi ed since the WPP provide a huge range of demographic indicators Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft Vol. 35, 2 (2010): 315-334 © Federal Institute for Population Research 2010 URL: www.comparativepopulationstudies.de DOI: 10.4232/10.CPoS-2010-08en URN: urn:nbn:de:bib-cpos-2010-08en0 • Marc Luy316 for current, former and expected future trends for every single country. Moreover, the WPP are prepared very carefully by using the best data available. The existence of such a comprehensive data source may easily conceal the fact that the estima- tion of demographic conditions and trends is a problematic and sophisticating task for most populations in the world, above all for those of developing countries. Most demographers and population researchers are merely used to the classical direct methods, which are based on sex- and age-specifi c counts of events like births or deaths and of the living population at risk. In most developing countries, however, such data are either lacking or severely defi cient to an extent that the common di- rect methods cannot be applied. Mainly during the 1970s and 1980s, demographers developed a number of spe- cifi c estimation methods to overcome these data problems. These specifi c estima- tion tools fall into two major categories: consistency checks with corresponding adjustment methods and indirect estimation techniques. The fi rst group comprises methods which are used for checking the completeness of registered data and for adjusting for typical underreporting in order to obtain less biased demographic measures. In the fi eld of mortality research, the most commonly used consistency checks/adjustment methods are different variants of the growth balance method (Brass 1975; Martin 1980; Gray 1986; Hill 1987; Bhat 2002; Hill/Queiroz 2004) and the inter-census estimation techniques (e.g. Preston/Bennett 1983). With indirect techniques, mortality estimates are derived from survey information on the survival experiences of close relatives or household members. For the estimation of infant and child mortality, the most frequent method is based on information about children ever born and children surviving (Sullivan 1972, Brass 1975, Trussell 1975, Preston/ Palloni 1977, Hill et al. 1983, Hill 1991), whereas the most prominent indirect method for the estimation of adult mortality is the orphanhood method, which is based on survey reports on parental survival. Different approaches have been suggested to convert the share of people with mother or father alive into life table survivorship ratios (Hill/Trussell 1977, Hill et al. 1983, Chackiel/Orellana 1985, Timæus 1991a/b, 1992, Timæus/Nunn 1997) or to use two sets of orphanhood data to estimate adult mortality for the time between the surveys (Zlotnik/Hill 1981, Timæus 1986). Other approaches for the indirect estimation of adult mortality are based on the survival of spouses (Hill/Trussell 1977) and siblings (Hill/Trussell 1977, Gakidou/King 2006, Obermeyer et al. 2010). Model life tables – like those by Coale and Demeny (Coale/ Demeny 1966, Coale et al. 1983), the United Nations (United Nations 1982a/b) or the INDEPTH network (Ngom/Bawah 2004) – are used to translate survival estimates for specifi c age segments derived from such indirect techniques into complete life tables and thus into life expectancy. After two decades of intensive discussion and derivation of these specifi c esti- mation tools, the use of consistency checks/adjustment methods and indirect es- timation techniques seems largely forgotten among those who are not forced to apply them. However, for the majority of developing countries these methods are still the main and often the only possible estimation tool, even for the most recent demographic data. This paper deals with the methods used by the United Nations to estimate life tables and thus life expectancy for their WPP. Since the fi nal estimates A Classifi cation of the Nature of Mortality Data • 317 are published in one table for all countries, it might seem that the numbers were directly comparable. The different nature of the underlying data and estimation methods, however, should be kept in mind before proceeding with too detailed in- terpretations. For the three most recent WPP, i.e. the 2004, 2006 and 2008 revisions, the United Nations published some brief information on the available data sources and the applied methods underlying the estimates of current and former conditions, which build the base of the projected future trends. In order to systematise the avail- able data sources and applied estimation techniques, we developed a fi ve-scale classifi cation of the nature of mortality data and assigned all countries into the cor- responding groups. Note that the nature of mortality data does not necessarily re- fl ect the accuracy of the data itself, although indirect estimates always include more uncertainties than direct estimates. Since in a number of cases the information pro- vided on the used data and methods is rather vague, our classifi cation contains some uncertainties as well. Furthermore, the information provided in this report is not comparable to the extensive publication of Lopez et al. (2002) who supplied a detailed overview of the specifi c data sources and methods used for constructing life tables for 191 WHO member states referring to the situation in the late 1990s. As will be shown in the present paper, developing countries most recently experience considerable improvements in the quality of the available mortality data. 2 A classifi cation of the nature of mortality data The creation of a classifi cation of the nature of mortality data has been inspired by similar works of Lopez et al. (2006) and Wilmoth (2007) who both described fi ve categories of data availability but confi ned their work to a general description rather than providing the corresponding classifi cation for each country. The criteria used for assigning countries to one of these groups differ slightly from one another and also from the criteria we used, and as such, the classifi cation presented in this pa- per differs from both as well. Our classifi cation is based on three major sources on information regarding the nature of mortality data, the analytical reports of the 2004 and 2006 revisions of United Nations’ WPP (United Nations 2006, 2008) and the methods and data descriptions of the “Global Burden of Disease” (GBD) (Lopez et al. 2006). The 2008 revision of United Nations’ WPP (United Nations 2010) has not been used because of lacking comparability to the GBD classifi cation as will be dis- cussed in more detail in the concluding remarks. Both the WPP and the GBD provide some – though differently structured – information on data sources and methods underlying the corresponding mortality estimates. Unfortunately, the two sources cannot be matched directly since the GBD published some of the information only for groups of countries following from the classifi cation system of the World Health Organisation (WHO) which differs from the corresponding regional classifi cation system of the United Nations. However, since Lopez et al. (2006) estimated their own life tables for the most part independently from those of the United Nations, the GBD was considered a useful tool for checking our interpretation of the partly vague information provided by the United Nations. • Marc Luy318 Our fi ve-group classifi cation is based on the following criteria: Group I contains countries with complete and reliable vital registration based • on census data. The age- and sex-specifi c mortality rates as a basis for the life tables are calculated directly and unadjusted from the offi cially regis- tered data. Group II contains countries for which the calculation or derivation of age- • and sex-specifi c death rates is still possible. However, some adjustments of the registered data or the use of additional data sources are necessary. This group contains countries of three kinds of data characteristics: (i) countries whose vital registration and census data show some defi ciencies making adjustments of the numbers of deaths or the living population necessary, (ii) countries with current death registration in which former census data or life tables allow to derive estimates for the current age and sex composition of the living population from specifi c projection or modelling, and (iii) countries with existing death registers but whose data on infant and/or child mortal- ity are based on other data sources, i.e. indirect estimates from survey or census data. Group III contains countries whose life tables are not based on age- and sex-• specifi c death rates but on two separated estimates, one for child mortal- ity and one for a global level of adult mortality. Estimates of child mortality are based on data from registers, censuses, DHS and PAPCHILD surveys or surveillance systems. Regarding the estimation of adult mortality levels, the countries assigned to this group fall into two broad categories: (i) countries whose estimates for adult mortality are based on former life tables dating back too far to derive current age- and sex-specifi c death rates by projection techniques but allow deriving estimates for a global level of adult mortality, and (ii) countries with estimates of adult mortality levels from indirect esti- mation techniques based on information about surviving parents or siblings from survey data. In most countries of this group, the fi nal estimates for life expectancy are derived from model life table systems or are adopted from alternative sources like WHO or ESCWA estimates. Group IV contains countries whose estimates for life expectancy are solely • based on direct or indirect estimates of child mortality, using model life table systems for the derivation of the corresponding life table for the total popu- lation. As in group III, estimates of child mortality are based on data from registers, censuses, DHS and PAPCHILD surveys or surveillance systems. In some cases, UNICEF and WHO estimates were considered as well. Group V contains countries without any useful information on prevailing • mortality conditions. In these cases, the estimated life expectancy is as- sumed to be similar to those of neighbouring countries with comparable socio-economic conditions, usually belonging to group IV. In practice, classifying countries into groups I or V was mostly unproblematic. However, the classifi cation of countries into the groups II, III and IV was a sophisti- cated task given the information provided on the data sources and methods for life table derivation. Nevertheless, using all available information from the WPP 2004 and WPP 2006 analytical reports and the GBD, we kept our classifi cation criteria as clearly as possible. First, we classifi ed the countries for the WPP 2004 since the years of data collection are roughly the same as for the GBD and therefore the information from the two sources could be combined. Our fi nal classifi cation of A Classifi cation of the Nature of Mortality Data • 319 countries into the fi ve groups for the WPP 2004 is mostly consistent with the ana- lytical reports of the United Nations (2006) and very close to the GBD classifi cation. The classifi cation for the WPP 2006 is based on the analytical reports of the United Nations (2008) and on our own classifi cation for the WPP 2004. Note that since both the WPP analytical reports and the GBD description are used for assigning countries into the groups of nature of mortality data, some classifi cations might not correspond to the description of one report on its own. Some examples might help explaining the diffi culties for, and the logic behind, our classifi cation. For the majority of the African countries the WPP analytical reports describe the source for life expectancy estimates as “derived from estimates on infant and child mortality by assuming that the age pattern of mortality conforms to the […] model of the […] Model Life Tables”. In most cases this statement is supplemented by “the demographic impact of AIDS has been factored into the mortality estimates”. In cases where no further information is given other than that the estimates are based on infant and child mortality, countries were assigned to group IV. For some coun- tries like Bhutan, the WPP analytical report additionally included the information that some older (offi cial) life table estimates were considered. However, it remains unclear if these additional sources were used for estimating adult mortality levels or if they were used to test and adjust the estimates for child mortality. For classifying such countries we incorporated the information from the GBD, where for the main WHO regions the number of countries is listed by their fi ve classes of mortality data. If the GBD classifi cation indicates that the country should belong to group III we as- signed this country to group III as well. For other countries the data description of the WPP 2004 analytical report leaves open whether they belong to group II or to group III. If the GBD team could not reconstruct age-and sex-specifi c rates from the data of such countries, they were assigned to group III, as long as the WPP 2004 analytical report did not contain any explicit remark on age- and sex-specifi c data availability. Consequently, when the description in the WPP 2006 analytical report did not change, the country remained in data quality group III, although the analytical report itself might indicate mortality data of quality group II. Algeria is an example for such a classifi cation. Furthermore, there are other countries theoretically fulfi lling the criteria for group II (or even group I) but where the fi nal estimates of the United Nations are based on model life tables. Such countries were classifi ed into group III, since obvi- ously the data quality is insuffi cient to derive age- and sex-specifi c death rates. Tu- nisia is such a country; it has regular censuses and offi cial estimates for infant and child mortality as well as for life expectancy provided by INS Tunisia (see also Vallin/ Locoh 2001). However, the life expectancy in Tunisia reported in the WPP is based on an underlying age pattern of mortality that conforms to the East model of the Coale-Demeny model life table system. Bolivia and South Africa are further exam- ples of countries where the theoretically available data contradict our classifi cation. In Bolivia, the national statistical offi ce publishes offi cial life tables while the United Nations based their estimates for the WPP on indirect estimations. The same holds for South Africa, where works of local demographers suggest that the available data could provide adjusted age- and sex-specifi c death rates and that the application of • Marc Luy320 model life tables might not be necessary (Dorrington et al. 2006). However, since the United Nations choose the data for their WPP very carefully, there seem to be justifi ed reservations with regard to the quality of such sources. Finally, there are countries such as Morocco, for which, according to the WPP analytical reports, no information on adult mortality exist. However, regarding the trend of the age pattern of mortality the United Nations assumed for the WPP 2006 a convergence from the South to the East level of the Coale-Demeny model life ta- bles, whereas all former, current and future estimates of the WPP 2004 were based solely on the West level. This indicates that at least some information on adult mor- tality patterns must have been available, and thus, Morocco has been classifi ed into group III. Furthermore, classifying Morocco into group III was necessary in order to create fi gures in accordance with the GBD classifi cation for countries of the WHO region ‘Middle East and North Africa’. 3 A worldwide overview of the nature of mortality data The distribution of countries with more than 100,000 inhabitants into the fi ve groups of nature of mortality data underlying the WPP 2004 and WPP 2006 is shown in ta- bles 1 and 2 for the world’s major areas, as well as in fi gures 1 and 2 for each country. Furthermore, the country-specifi c classifi cation for the WPP 2004 and the WPP 2006 and the model life table systems used for the WPP 2006 can be found in the appen- dix of this paper. Table 1 shows that for the life expectancy estimates of the United Nations’ WPP 2004 the countries are almost evenly distributed among groups I-IV. There are only two countries without any data on mortality, Western Sahara and Guinea-Bissau. The group to which most countries are assigned to is group IV with 52 countries. 42 of these countries are from Africa, seven from Asia, two from Latin America and one from Oceania. Group III contains 42 countries, dominated by 18 from Asia. The other 24 countries of this group are distributed evenly across Africa, Oceania, and Latin America and the Caribbean. Group II consisting of 47 countries mainly comprises countries from Latin America and the Caribbean (21) and Asia (16). Further, eight European countries and one country from each Africa and Oce- ania also belong to this group. The group with the highest quality of mortality data contains 49 countries out of which most are from Europe (31) but also from Asia (9), Latin America and the Caribbean, furthermore from the United States and Canada, Australia and New Zealand as well as Mauritius (see also Fig. 1). A comparison between the WPP 2004 and WPP 2006 reveals some dynamics in the nature of mortality data underlying the most recent life expectancy estimates, mainly in terms of having more detailed data available (see Tab. 1 and Tab. 2). The most signifi cant changes are shifts from group IV to group III which mainly occurred in western African countries (Burkina Faso, Gambia, Mali, Mauritania, Niger and Senegal), but also in São Tomé and Príncipe, Libya, South Africa, Bhutan, Iran, Saudi Arabia, Yemen, and Saint Vincent and the Grenadines. Three countries moved in our classifi cation from group III to group II, the Democratic People’s Republic of Korea, Sri Lanka and Kuwait. However, according to the analytical reports of WPP 2004 and A Classifi cation of the Nature of Mortality Data • 321 WPP 2006, there are also shifts in the other direction. Two countries changed from group I to group II (Bahrain and Ukraine), one from group II to group III (Maldives) and another one from group III to group IV (Lebanon). These shifts towards less detailed and less accurate mortality data are based on the WPP analytical reports. Tab. 1: Classifi cation of the nature of mortality data underlying the estimation of life expectancy in United Nations’ World Population Prospects 2004 for the world’s major areas Group of nature of mortality data I II III IV V WORLD 49 47 42 52 2 Africa 1 1 8 42 2 Asia 9 16 18 7 0 Europe 31 8 0 0 0 Latin America and the Caribbean 4 21 8 2 0 Northern America 2 0 0 0 0 Oceania 2 1 8 1 0 Source: own reconstruction based on information provided by the United Nations (2006) and Lopez et al. (2006) Fig. 1: Classifi cation of the nature of mortality data underlying the estimation of life expectancy in United Nations’ World Population Prospects 2004 Data: own reconstruction based on information provided by the United Nations (2006) and Lopez et al. (2006) • Marc Luy322 In Bahrain, for instance, adjustments had to be made for infant and child mortality for the WPP 2006, which was not necessary according to the WPP 2004 analytical report. The Maldives changed from group II to group III because the estimates for life expectancy of the WPP 2004 had been based on offi cial estimates, whereas the Tab. 2: Classifi cation of the nature of mortality data underlying the estimation of life expectancy in United Nations’ World Population Prospects 2006 for the world’s major areas Group of nature of mortality data I II III IV V WORLD 47 51 53 39 2 Africa 1 1 17 33 2 Asia 8 19 19 4 0 Europe 30 9 0 0 0 Latin America and the Caribbean 4 21 9 1 0 Northern America 2 0 0 0 0 Oceania 2 1 8 1 0 Source: own reconstruction based on information provided by the United Nations (2006, 2008) and Lopez et al. (2006) Fig. 2: Classifi cation of the nature of mortality data underlying the estimation of life expectancy in United Nations’ World Population Prospects 2006 Data: own reconstruction based on information provided by the United Nations (2006, 2008) and Lopez et al. (2006) A Classifi cation of the Nature of Mortality Data • 323 WPP 2006 estimates used Coale-Demeny model life tables. In sum, regarding the nature of mortality data underlying the life expectancy estimates of the WPP 2006, most countries belong to group III (53), followed by group II (51), group I (47) and group IV (39). No improvements took place in Western Sahara and Guinea-Bissau, so both countries still belong to group V for the WPP 2006 estimates as they already did for the WPP 2004 (see also Fig. 2). Looking at the distribution of countries into the fi ve groups of nature of mortality data for the world’s major areas reveals that most African countries belong to group IV, despite the remarkable shifts to group III. Most Asian countries belong to the groups II and III, although there are still four countries belonging to group IV (Leba- non, Nepal, Indonesia and Laos). Most European countries produce more detailed and more accurate mortality data. However, there are still nine countries from east- ern and south-eastern Europe where an adjustment of age- and sex-specifi c death rates is necessary (Belarus, Moldova, Ukraine, Estonia, Lithuania, Albania, Bosnia and Herzegovina, Malta, TFYR Macedonia). Most countries from Latin America and the Caribbean belong to group II, nine to group III, four to group I (Argentina, Chile, Cuba, Netherlands Antilles), and with Belize one to group IV. Apart from Australia and New Zealand, most countries from Oceania belong to group III. There is only one country each in group II (French Polynesia) and group IV (Federal States of Micronesia). Regarding the groups of nature of mortality data of the WPP 2006, it has already been said that group I contains developed high-income countries from Europe, America and Asia (see Tab. 2 and Fig. 2). Most of the 51 countries belonging to group II are from Latin America and the Caribbean (21) and Asia (19), complemented by the nine European countries mentioned above as well as Réunion and French Polynesia from Africa and Oceania, respectively. The 53 countries of group III are mainly countries from Asia (19) and Africa (17), complemented by nine from Latin America and the Caribbean and eight from Oceania. With 33 out of 39 countries, Africa dominates group IV. The other six countries of this group are Lebanon, Nepal, Indonesia, Laos (Asia), Belize (Latin America and the Caribbean) and the Federal States of Micronesia (Oceania). Finally, we looked at the specifi c estimation tools that were used for estimates of life expectancy of the WPP 2006 in cases where the available mortality data was insuffi cient to produce direct age- and sex-specifi c death rates. In most cases, the indirect estimates of infant and child mortality are based on the number of chil- dren ever born and children surviving obtained from survey or census information, or from maternity history and the number of births during the twelve preceding months from the same sources of information. Regarding the estimation of adult mortality, the orphanhood method was explicitly mentioned in 13 cases (Burkina Faso, Gambia, Mali, Mauritania, Niger, Senegal, Dominican Republic, Bolivia, Papua New Guinea, Solomon Island, Vanuatu, Samoa, Tonga), the growth balance method and its variants were given ten times (Haiti, El Salvador, Honduras, Panama, Brazil, Colombia, Ecuador, Paraguay, Peru, Venezuela). At least for the latter, the number of applications is probably higher since in most cases the only information was that some adjustment had been done without any specifi cation of the used method. • Marc Luy324 In order to overcome the lack of data in modelling life tables on some informa- tion about child – and in some cases additionally on adult – mortality levels, for 79 out of the 92 countries belonging to quality groups III and IV model life tables were used (see Appendix). The most commonly used model life tables are those from the Coale-Demeny model life table system, in particular the West pattern. The Coale- Demeny West pattern provides the background for 27 life table estimates, whereas the North pattern was used for 20, the South pattern for eight and the East pattern for six country estimates. For twelve countries, the United Nations model life table system was applied, dominated by the Far Eastern pattern which was used for the estimates for nine countries, while the Latin American, the South Asian and the General pattern each were used once. For six countries in western Africa (Burkina Faso, Gambia, Mali, Mauritania, Niger, Senegal) the traditional Coale-Demeny mod- el life table system was replaced by Pattern 1 from the recently published INDEPTH model life tables. Interestingly, there are some other cases where the model life table system used was changed between the WPP 2004 and WPP 2006 which also changed the as- sumed pattern of age-specifi c mortality indicating the existence of new informa- tion regarding former and current mortality trends. The model life tables have been changed for Laos from Coale-Demeny North to Coale-Demeny West, for Barbados, Cambodia and Afghanistan from Coale-Demeny South to Coale-Demeny West, for Tanzania from Coale-Demeny South to Coale-Demeny North, for Chad from Coale- Demeny North to Coale-Demeny South, for Morocco from Coale-Demeny West to Coale-Demeny East, for Belize from UN Latin American to UN General, and fi nally for the Solomon Islands from UN Far Eastern to Coale-Demeny West. 4 Concluding remarks This paper makes an attempt at creating a fi ve-scale classifi cation of nature of mor- tality data for all countries of the world with more than 100,000 inhabitants, referring to the data and methods used by the United Nations for their WPP, namely the 2004 and 2006 revisions. These are the fi rst two WPP for which at least some informa- tion about the specifi c data and methods is provided by analytical reports. Besides describing the classifi cation criteria we also assessed the worldwide situation of mortality data availability and the changes which occurred between the two WPP. The most important change was the upgrade of many – mainly African – countries from group IV to group III. However, Africa and to some extent south-eastern Asia are still the areas with most incomplete and probably least reliable data about mor- tality levels and trends. Technically, the major change between WPP 2004 and WPP 2006 is the extension of the model life table systems used to the INDEPTH model life tables, and the use of data from the Human Mortality Database (HMD) instead of material from offi cial sources for some countries from eastern and south-eastern Europe (Bulgaria, Russian Federation, Slovakia, Ukraine). The most recently published 2008 revision of United Nations’ WPP indicates that with regard to the nature and accuracy of mortality data the situation has fur- A Classifi cation of the Nature of Mortality Data • 325 ther improved compared to the WPP 2006. This mainly concerns some African as well as eastern and western Asian regions. On the other side, the most recent life table estimates for the Russian Federation and Ukraine incorporated adjustments to infant mortality, and in those for the U.S. the demographic impact of AIDS has been factored. These adjustments were not done in the estimates of the WPP 2006 according to the corresponding analytical report. In this paper we did not include the WPP 2008 in our analysis since many descriptions in the actual analytical report are too vague for classifi cation and would need additional information. Since there have been some obvious changes compared to the WPP 2004 and WPP 2006, the GBD 2006 is not suffi ciently up-to-date. Nevertheless, it is interesting to note that the United Nations stopped using the INDEPTH model life tables for the WPP 2008 life expectancy estimates and substituted them by the Timæus Sahelian mortality pattern (Timæus 1999). Further changes in the used model life table system were made for Sierra Leone (from Coale-Demeny South to the Timæus Western and East- ern Africa mortality pattern), Liberia (from Coale-Demeny West to Coale-Demeny South) and Iraq (from Coale-Demeny East to Coale-Demeny West and the Brass General Standard). In the fi nal count, the developed classifi cation for the WPP 2004 and WPP 2006 might not be perfectly precise for every country. In particular, the differentiation between groups II and III was diffi cult in some cases. Furthermore, the classifi cation does not provide a detailed overview of the specifi c methods and approaches used by the United Nations Population Division. For instance, the new data from the DHS/ RHS/Arab League etc. birth histories have led to the production of a whole new set of fi gures based on birth history analysis. Detailed insights into the specifi c meth- ods have been omitted because the intention of this paper was to provide a crude and descriptive overview of the situation of mortality statistics underlying the WPP 2004 and WPP 2006. However, the paper should by no means be understood as criticism on the work of the United Nations Population Division and their estimates and projections of life expectancy. The lack of data in most developing countries is not the fault of the United Nations and their efforts in producing the most reli- able estimates of life expectancy deserve high credit. Above all the United Nations Population Division’s work factoring the demographic impact of AIDS is particularly diffi cult and very well done. Likewise, the goal of this overview was neither to discuss the processes or mech- anisms by which countries moved up in the classifi cation, nor to provide a detailed assessment of the state of vital registration in each country. This would require a careful analysis and much more extensive literature research for each country. Nevertheless, the contents of this paper should be of interest for politicians and scientists using the United Nations WPP as well as for scholars who teach and learn about indirect demographic estimation techniques. For the latter, it is especially in- teresting to see in which countries these methods are still applied, in order to realise the important role which indirect methods and model life tables still play in recent demographic estimates of the world. The author thanks Werner Richter and Barbara Müller for language editing. • Marc Luy326 References Bhat, P.N. Mari, 2002: General growth balance method: a reformulation for populations open to migration. In: Population Studies 56,1: 23-34 Brass, William, 1975: Methods for estimating fertility and mortality from limited and defective data. Chapel Hill: University of North Carolina Chackiel, Juan; Orellana, Hernán, 1985: Adult female mortality trends from retrospec- tive questions about maternal orphanhood included in censuses and surveys. In: International Union for the Scientifi c Study of Population (ed.): IUSSP International Population Conference, Florence 1985, 4. Liège: IUSSP: 39-51 Coale, Ansley J.; Demeny, Paul, 1966: Regional model life tables and stable populations. Princeton: Princeton University Press Coale, Ansley J.; Demeny, Paul; Vaughan, Barbara, 1983: Regional model life tables and stable populations. Second edition. New York et al.: Academic Press Dorrington, Rob; Timæus Ian M.; Gregson, Simon, 2006: Adult mortality in southern Africa using deaths reported by households: some methodological issues and results. Paper presented at the PAA 2006 Annual Meeting. Los Angeles, March 2006 Gakidou, Emmanuela; King, Gary, 2006: Death by survey: estimating adult mortality without selection bias from sibling survival data. In: Demography 43,3: 569-585 Gray, Alan, 1986: Sectional growth balance analysis for non-stable closed populations. In: Population Studies 40,3: 425-436 Hill, Kenneth, 1987: Estimating census and death registration completeness. In: Asian and Pacifi c Population Forum 1,3: 8-13 Hill, Kenneth, 1991: Approaches to the measurement of childhood mortality: a compara- tive review. In: Population Index 57,3: 368-382 Hill, Kenneth; Queiroz, Bernardo, 2004: Adjusting general growth balance method for migration. Paper presented at the AMDC Meeting. Berkley, July 2004 Hill, Kenneth; Trussell, James, 1977: Further developments in indirect mortality estima- tion. In: Population Studies 31,2: 313-334 Hill, Kenneth; Zlotnik, Hania; Trussell, James, 1983: Manual X: Indirect techniques for demographic estimation. New York: United Nations Lopez, Alan D.; Ahmad, Omar B.; Guillot, Michel; Ferguson, Brodie D.; Salomon, Joshua A.; Murray, Christopher J.L.; Hill, Kenneth, 2002: World mortality in 2000: life tables for 191 countries. Geneva: WHO Lopez, Alan D.; Mathers, Colin D.; Ezzati, Majid; Jamison, Dean T.; Murray, Christopher J.L., 2006: Global burden of disease and risk factors. New York: Oxford University Press, World Bank Martin, Linda G., 1980: A modifi cation for use in destabilized populations of Brass’s technique for estimating completeness of death registration. In: Population Studies 34,2: 381-395 Ngom, Pierre; Bawah, Ayaga A., 2004: INDEPTH model life tables for Sub-Saharan Af- rica. Aldershot/Burlington: Ashgate Obermeyer, Ziad; Rajaratnam, Julie Knoll; Park, Chang H.; Gakidou, Emmanuela; Hogan, Margaret C.; Lopez, Alan D.; Murray, Christopher J. L., 2010: Measuring adult mortal- ity using sibling survival: a new analytical method and new results for 44 countries, 1974-2006. In: PLOS Medicine 7,4: e1000260 A Classifi cation of the Nature of Mortality Data • 327 Preston, Samuel H.; Bennett, Neil G., 1983: A census-based method for estimating adult mortality. In: Population Studies 37,1: 91-104 Preston, Samuel H.; Palloni, Alberto, 1977: Fine-tuning Brass-type mortality estimates with data on ages of surviving children. In: Population Bulletin of the United Nations 10-1977: 72-91 Sullivan, Jeremiah M., 1972: Models for the estimation of probabilities of dying between birth and exact ages of early childhood. In: Population Studies 26,1: 79-97 Timæus, Ian M., 1986: An assessment of methods for estimating adult mortality from two sets of data on maternal orphanhood. In: Demography 23,3: 435-450 Timæus, Ian M., 1991a: Estimation of adult mortality from orphanhood before and since marriage. In: Population Studies 45,3: 455-472 Timæus, Ian M., 1991b: Estimation of adult mortality from orphanhood in adulthood. In: Demography 28,2: 213-227 Timæus, Ian M., 1992: Estimation of adult mortality from paternal orphanhood: a re- assessment and a new approach. In: Population Bulletin of the United Nations 33: 47-63 Timæus, Ian M., 1999: Notes on a series of life table estimates of mortality in the coun- tries of the Sub-Saharan Africa region. Unpublished manuscript prepared for the WHO Timæus, Ian M.; Nunn, Andrew J., 1997: Measurement of adult mortality in populations affected by AIDS: an assessment of the orphanhood method. In: Health Transitions Review 7, Supplement 2: 23-43 Trussell, T. James, 1975: A re-estimation of the multiplying factors for the Brass technique for determining childhood survivorship rates. In: Population Studies 29,1: 97-107 United Nations, 1982a: Model life tables for developing countries. New York: United Nations United Nations, 1982b: Unabridged model life tables corresponding to the new United Nations model life tables for developing countries. New York: United Nations United Nations, 2006: World population prospects. The 2004 revision. Volume III: Ana- lytical report. New York: United Nations United Nations, 2008: World population prospects. The 2006 revision. Volume III: Ana- lytical report. New York: United Nations United Nations, 2010: World population prospects. The 2008 revision. Volume III: Ana- lytical report. New York: United Nations Vallin, Jacques; Locoh, Thérèse, 2001: Population et développement en Tunisie. La mé- tamorphose. Tunis: Cérès Wilmoth, John R., 2007: The duration of life throughout the world. What do we know and how do we know it? Paper presented at the Max Planck Institute for Demographic Research, Rostock, May 2007 Zlotnik, Hania; Hill, Kenneth, 1981: The use of hypothetical cohorts in estimating demo- graphic parameters under conditions of changing fertility and mortality. Demography 18,1: 103-122 • Marc Luy328 A German translation of this authorised original article by the author is available under the title “Eine Klassifi kation der Mortalitätsdaten für die Schätzungen der United Nations’ World Popu- lation Prospects 2004 und 2006”, DOI 10.4232/10.CPoS-2010-08de or URN urn:nbn:de:bib-cpos- 2010-08de8, at http://www.comparativepopulationstudies.de. Dr. Marc Luy ( ). Austrian Academy of Sciences, Vienna Institute of Demography, A-1040 Vienna, Austria. E-Mail: mail@marcluy.eu A Classifi cation of the Nature of Mortality Data • 329 2004 2006 Model Life Table System Africa Eastern Africa Burundi IV IV Coale-Demeny North Comoros IV IV Coale-Demeny West Djibouti IV IV Coale-Demeny West Eritrea IV IV UN Far East Ethiopia IV IV Coale-Demeny North Kenya IV IV Coale-Demeny North Madagascar IV IV Coale-Demeny North Malawi III III Coale-Demeny South Mauritius I I ----- Mozambique III III Coale-Demeny North Réunion II II ----- Rwanda III III ----- Somalia IV IV Coale-Demeny North Uganda IV IV Coale-Demeny North United Republic of Tanzania IV IV Coale-Demeny North Zambia IV IV Coale-Demeny North Zimbabwe IV IV Coale-Demeny North Middle Africa Angola IV IV Coale-Demeny North Cameroon IV IV Coale-Demeny North Central African Republic IV IV Coale-Demeny North Chad IV IV Coale-Demeny South Congo IV IV Coale-Demeny West Democr. Rep. of the Congo IV IV Coale-Demeny North Equatorial Guinea IV IV Coale-Demeny North Gabon IV IV Coale-Demeny North Sao Tome and Principe IV III Coale-Demeny South Northern Africa Algeria III III ----- Egypt III III Coale-Demeny East Libyan Arab Jamahiriya IV III UN Far East Morocco III III Coale-Demeny East Sudan IV IV Coale-Demeny North Tunisia III III Coale-Demeny East Western Sahara V V ----- Appendix Classifi cation of the nature of mortality data and used Model Life Table System underlying the data for the estimation of life expectancy in United Nations’ World Population Prospects 2004 and 2006 for each country • Marc Luy330 2004 2006 Model Life Table System Southern Africa Botswana IV IV Coale-Demeny West Lesotho IV IV Coale-Demeny West Namibia III III Coale-Demeny West South Africa IV III UN Far East Swaziland IV IV Coale-Demeny West Western Africa Benin IV IV Coale-Demeny South Burkina Faso IV III INDEPTH 1 Cape Verde IV IV Coale-Demeny West Côte d'Ivoire IV IV Coale-Demeny South Gambia IV III INDEPTH 1 Ghana IV IV Coale-Demeny North Guinea IV IV Coale-Demeny South Guinea-Bissau V V ----- Liberia IV IV Coale-Demeny West Mali IV III INDEPTH 1 Mauritania IV III INDEPTH 1 Niger IV III INDEPTH 1 Nigeria IV IV Coale-Demeny North Senegal IV III INDEPTH 1 Sierra Leone IV IV Coale-Demeny South Togo IV IV Coale-Demeny South Asia Eastern Asia China III III ----- China, Hong Kong Spec. Adm. R. I I ----- China, Macao Spec. Adm. Region I I ----- Democr. People's Rep. of Korea III II ----- Japan I I ----- Mongolia III III ----- Republic of Korea II II ----- South-central Asia Afghanistan III III Coale-Demeny West Bangladesh III III Coale-Demeny West Bhutan IV III Coale-Demeny North India II II ----- Iran (Islamic Republic of) IV III Coale-Demeny East Kazakhstan II II ----- Kyrgyzstan II II ----- A Classifi cation of the Nature of Mortality Data • 331 2004 2006 Model Life Table System Maldives II III Coale-Demeny West Nepal IV IV Coale-Demeny West Pakistan III III UN South Asia Sri Lanka III II ----- Tajikistan II II ----- Turkmenistan II II ----- Uzbekistan II II ----- South-eastern Asia Brunei Darussalam II II ----- Cambodia III III Coale-Demeny West Democr. Rep. of Timor-Leste III III Coale-Demeny West Indonesia IV IV Coale-Demeny West Lao People's Democratic Republic IV IV Coale-Demeny West Malaysia I I ----- Myanmar III III UN Latin America Philippines III III ----- Singapore I I ----- Thailand II II ----- Viet Nam II II ----- Western Asia Armenia II II ----- Azerbaijan II II ----- Bahrain I II ----- Cyprus I I ----- Georgia II II ----- Iraq III III Coale-Demeny East Israel I I ----- Jordan III III ----- Kuwait III II ----- Lebanon III IV Coale-Demeny West Occupied Palestinian Territory III III ----- Oman I I ----- Qatar II II ----- Saudi Arabia IV III Coale-Demeny West Syrian Arab Republic III III Coale-Demeny West Turkey III III Coale-Demeny East United Arab Emirates II II Coale-Demeny West Yemen IV III Coale-Demeny West • Marc Luy332 2004 2006 Model Life Table System Europe Eastern Europe Belarus II II ----- Bulgaria I I ----- Czech Republic I I ----- Hungary I I ----- Poland I I ----- Republic of Moldova II II ----- Romania I I ----- Russian Federation I I ----- Slovakia I I ----- Ukraine I II ----- Northern Europe Channel Islands I I ----- Denmark I I ----- Estonia II II ----- Finland I I ----- Iceland I I ----- Ireland I I ----- Latvia I I ----- Lithuania II II ----- Norway I I ----- Sweden I I ----- United Kingdom I I ----- Southern Europe Albania II II ----- Bosnia and Herzegovina II II ----- Croatia I I ----- Greece I I ----- Italy I I ----- Malta II II ----- Portugal I I ----- Serbia and Montenegro I I ----- Slovenia I I ----- Spain I I ----- TFYR Macedonia II II ----- Western Europe Austria I I ----- Belgium I I ----- France I I ----- A Classifi cation of the Nature of Mortality Data • 333 2004 2006 Model Life Table System Germany I I ----- Luxembourg I I ----- Netherlands I I ----- Switzerland I I ----- Latin America and the Caribbean Caribbean Bahamas II II ----- Barbados III III Coale-Demeny West Cuba I I ----- Dominican Republic III III ----- Guadeloupe II II ----- Haiti III III ----- Jamaica II II ----- Martinique II II ----- Netherlands Antilles I I ----- Puerto Rico II II ----- Saint Lucia II II ----- Saint Vincent and the Grenadines IV III Coale-Demeny West Trinidad and Tobago III III ----- United States Virgin Islands III III UN Far East Central America Belize IV IV UN General Costa Rica II II ----- El Salvador II II ----- Guatemala II II ----- Honduras II II ----- Mexico II II ----- Nicaragua II II ----- Panama II II ----- South America Argentina I I ----- Bolivia III III ----- Brazil II II ----- Chile I I ----- Colombia II II ----- Ecuador II II ----- French Guiana III III Coale-Demeny West Guyana III III ----- Paraguay II II ----- Peru II II ----- • Marc Luy334 2004 2006 Model Life Table System Suriname II II ----- Uruguay II II ----- Venezuela II II ----- Northern America Canada I I ----- United States of America I I ----- Oceania Australia/New Zealand Australia I I ----- New Zealand I I ----- Melanesia Fiji III III UN Far East New Caledonia III III Coale-Demeny West Papua New Guinea III III UN Far East Solomon Islands III III Coale-Demeny West Vanuatu III III UN Far East Micronesia Guam III III ----- Micronesia (Federated States of) IV IV Coale-Demeny West Polynesia French Polynesia II II ----- Samoa III III UN Far East Tonga III III UN Far East Note: ----- not applicable Source: own reconstruction based on information provided by the United Nations (2006, 2008) and Lopez et al. (2006). Published by / Herausgegeben von Prof. Dr. Norbert F. Schneider Layout and print: Federal Institute for Population Research, Wiesbaden (Germany) Managing Editor / Redaktion Frank Swiaczny Copy Editor / Schlußredaktion Dr. Evelyn Grünheid Scientifi c Advisory Board / Wissenschaftlicher Beirat Jürgen Dorbritz (Wiesbaden) Paul Gans (Mannheim) Johannes Huinink (Bremen) Dirk J. van de Kaa (Den Haag) Marc Luy (Wien) Notburga Ott (Bochum) Peter Preisendörfer (Mainz) Comparative Population Studies – Zeitschrift für Bevölkerungswissenschaft www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet) Board of Reviewers / Gutachterbeirat Martin Abraham (Erlangen) Laura Bernardi (Lausanne) Hansjörg Bucher (Bonn) Claudia Diehl (Göttingen) Andreas Diekmann (Zürich) Gabriele Doblhammer-Reiter (Rostock) Henriette Engelhardt-Wölfl er (Bamberg) E.-Jürgen Flöthmann (Bielefeld) Alexia Fürnkranz-Prskawetz (Wien) Beat Fux (Zürich) Joshua Goldstein (Rostock) Karsten Hank (Mannheim) Sonja Haug (Regensburg) Franz-Josef Kemper (Berlin) Hans-Peter Kohler (Philadelphia) Michaela Kreyenfeld (Rostock) Aart C. Liefbroer (Den Haag) Kurt Lüscher (Konstanz) Dimiter Philipov (Wien) Tomáš Sobotka (Wien) Heike Trappe (Rostock) © Federal Institute for Population Research 2010 – All rights reserved Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext Feiler-Fuchs-B Schreibmaschinentext