CENSUS

ii. In Afghanistan

The first national census of Afghanistan was not conducted until 1358 Š./1979. Before that Afghanistan was among the few countries in the world that had never undertaken such a survey, and, as a result, it suffered all the socioeconomic uncertainties that might be expected from such a lack. Although the census of 1358 Š./1979 by no means eliminated all these uncertain­ties (see below), it nevertheless marked the beginning in Afghanistan of scientific reporting of the population.

Even though it was a long time before this first Afghan census was undertaken, the idea of such a survey had nevertheless taken root fairly early, in the reign of Šēr-ʿAlī Khan (1285-96/1868-79), when gradual suppression of tax farming in favor of direct collection of taxes by government officials made it imperative for the administration to know the number of taxable households. A first count of the population of Kabul was carried out in 1293/1876, and the ethnic affiliations of the residents were recorded. Only the total figures have survived, however (Gazetteer of Afghanistan VI, p. 333). Unfortunately, not even the totals are known from a census that is supposed to have taken place in the Kōhdāman and Kōhestān districts of Kabul province at about the same period (Iftikhar-ud-Din, p. 133), nor is it known whether comparable efforts were undertaken elsewhere in Afghanistan. It is, further­more, impossible to determine the degree of reliability of these pioneering surveys, the results of which must thus be considered merely as indicators.

The British occupation of southeastern Afghanistan in 1295-98/1878-81 was the occasion for several local surveys conducted with a relatively high degree of precision, in particular the census carried out at Qan­dahār in 1880, which provided data arranged by quarter and by sex (Gazetteer of Afghanistan V, pp. 242-46). During the same interval Doctors Brereton and Tully, successively in charge of the British dispensary at Qandahār, launched an epidemiological study and inquiry into the mortality rate, both unprecedented in Afghanistan (Gazetteer of Afghanistan V, pp. 255-56; Tully, Sanitary and Topographic Report on the City of Kandahar, Quetta, 1 June 1881, IOR L/P & S/7/29/1097-1107).

Throughout his long reign ʿAbd-al-Raḥmān Khan (1297-1319/1880-1901) was also eager to gain more information about the population of his kingdom, for both fiscal and military reasons. Especially during the years 1307-09/1890-91 he extended the practice of conducting simple enumerations to new regions: Afghan Turkestan (Mazar Newsletter, 27 January 1890, IOR L/P & S/7/59/820; Trans-Frontier Journal, April 1890, IOR L/P & S/7/60/13), Badaḵšān (Kabul Newsletter, 13 May 1891, IOR L/P & S/7/63/299), Laḡmān (Pesha­war Confidential Diary, 21 August 1891, IOR L/P & S/7/63/1286), and Nangrahār (Peshawar Confidential Diary, 23 September 1891, IOR L/P & S/7/64/228). In the Ḵōst area of Paktīā a simple count of houses was carried out (Kabul Newsletter, 20 September 1887, IOR L/P & S/7/51/646). In Herat the population was coun­ted on two occasions, in 1308/1891 (Trans-Frontier Journal, May 1891, IOR L/P & S/7/63/495) and 1316/1898 (Herat Newsletter, 11 August 1898, NAI Foreign Department, Secret F, December 1898, 47). The popu­lation of the city of Qandahār was even counted three times, in 1305/1887 (Kandahar Newsletter, 7 October 1887, IOR L/P & S/7/51/789), 1309/1891 (Kandahar Newsletter, 10 August 1891, IOR L/P & S/7/63/1295), and 1315/1898 (Kandahar Newsletter, 7 January 1898, NAT Foreign Department, Secret F, February 1898, 1304). Such repetition after short intervals is an indi­cator of the poor quality of these early enumerations, which were conducted primarily by ḥākems (governors), qāżīs (judges), ʿolamāʾ, and mollās; Kakar, p. 182). Despite the amir’s efforts to achieve national coverage, many regions remained outside the scope of operations (Kabul Newsletter, 13 May, 1891, IOR L/P & S/7/63/299). Furthermore, as the only result that is still extant is that of the 1309/1891 census of Qandahār, the intense activity generated by ʿAbd-al-Raḥmān Khan’s policy contributed very little overall to the demographic history of Afghanistan.

In the same period the members of the Afghan Boundary Commission (1302-03/1884-86) undertook to count the populations of the regions that they traversed (see boundaries iii. boundaries of afghan­istan). As they were unable to produce full-fledged census reports, they had to be content with estimates, which varied in precision according to the locality. The majority of these estimates have been published in Gazetteer of Afghanistan, though, regrettably, the two most detailed were not included: those for the province of Herat (Muhammad Takki Khan, Report on the City and Province of Herat, IOR L/P & S/7/49/1175-1270) and the Hazārajāt region (Muhammad Hussain Khan, Report on the Hazarajat, IOR L/P & S/7/49/415-37). This collection of regional demographic data remained unmatched until the second half of the 14th/20th century.

The active interest in census taking that characterized Afghan rulers during the last quarter of the 13th/19th century was not apparent under their immediate suc­cessors. The difficulty of communications and the central government’s uncertain control over many regions of the country were major obstacles to efforts at overall enumeration. In addition, the ruling class showed an obvious lack of interest in statistical precision. It is significant in this respect that the reforms of Amān-Allāh (1337-48/1919-29) seem to have been entirely lacking in a demographic dimension: The only census that he planned was a count of domestic animals in 1302 Š./1923, for evasion of livestock taxes was widespread. From the beginning of the century, however, there survives a compilation of the first data collected on the nomadic population of the country, which, though certainly incomplete, is nevertheless highly instructive (completely published in Iftikhar-ud­-Din, appendix).

For several decades it has thus been possible to obtain only a very approximate, even intuitive, notion of Afghan demography, with all the risks that such an approach entails. For example, in the second decade of the century the bimonthly journal Serāj al-aḵbār published only a few years apart estimates of the Afghan population range from 5-6 million to 10 million inhabitants (Schinasi, p. 94). The margin of uncertainty still increased with time: In 1349 Š./1970 the population was reported as between 7 and 17 million inhabitants (L. Dupree, 1972, p. 31).

In the 1330 Š./1950s, however, at the urging of Afghan officials trained abroad, an interest in statistics in general and demography in particular was revived in Kabul. A directorate general of statistics was created in the ministry of interior, the forerunner of the present Central Statistics Office (Edāra-ye Markazī-e Eḥṣāʾīya), which has become an independent agency. The adop­tion of economic planning and the launching of the first five-year plan in 1325 Š./1956 also made the gathering of reliable demographic statistics increasingly urgent. It is against this background that the establishment of the Afghan civil register (taḏkera) must be understood. Under the new legislation each birth and death was to be registered with the local administrative authority within a seven-day period; in addition, all adults were to register immediately with the district administration and to receive civil-registration cards. In practice the compilation of the registers has turned into a long-term undertaking, which began in 1333 Š./1954 (Table 22) and was still unfinished twenty years later, when sixty out of 325 districts were still without such registers (Eighmy, 1976, p. 30). In theory the entire process was supposed to permit compilation of detailed demographic registers, which would be kept up to date, and it was planned that cumulative data for each province would also be published annually. All these hopes were disappointed, however. Births and deaths remained notoriously underreported, especially in the country­side, both because the requirement was resented as intolerable state interference in the intimate family sphere and because until 1350 Š./1971 there was a registration tax, which, though small (5 Afghanis), was sufficient to discourage registration. Among adults, on the other hand, it was primarily the men who registered and even then in proportions that varied considerably from one place to another.

It was recognized very early that only a genuine census conducted according to scientific criteria would permit elimination of all the uncertainties that sur­rounded the Afghan population. Nevertheless, insufficient financial means, qualified personnel, and politi­cal will at the highest levels meant that the undertaking was postponed from year to year. At first the govern­ment was content with several pilot surveys of an experimental kind. In 1333 Š./1954 a survey of house­holds in Kabul was carried out by the municipality on behalf of the World Health Organization, and average family size was estimated, on the basis of a random sample of families (Survey of Progress 1961-62, Kabul, 1342 Š./1963, pp. 29-30). Shortly afterward the first true censuses in Afghanistan were conducted experimentally in two suburban villages near the capital: Bagrāmī in Saraṭān 1338 Š./July 1959 (Nightingale and Spencer) and Deh Ḵodaydād in Ḥamal 1339 Š./April 1960 (Ministry of Planning, 1961). Several years later anew effort was undertaken with the Greater Kabul Census, accomplished in ten days in Asad 1344 Š./July 1965: This time 69,871 households, comprising 435,203 people, were counted (Ministry of Interior). A further series of sample surveys of employment and household expenditures, conducted by the Central Authority for Housing and Town Planning (Rīāsat-e ʿOmūrī-e Šahr­sāzī) in six major towns in the 1340s Š./1960s, also yielded significant demographic data.

In rural areas, however, the continuing inability of public authorities to furnish reliable demographic data forced a number of semipublic bodies to assemble the information they required. For example, the staff of the malaria-eradication program of the ministry of health compiled highly reliable demographic data on the malaria-stricken regions of the country. In 1350 Š./1971 the Paktya Development Authority organized a survey in Paktīā based on a sample of 1:50: Fifty-four of 2,730 villages, a population of 2,783 households and 24,809 people, were surveyed completely (Gerken). Several village censuses carried out by university re­searchers in different regions of the country must also be mentioned (Hahn; Toepfer).

Such relatively numerous local surveys, whether exhaustive or based on samples, have, however, con­tributed little to dispelling the obscurity that continues to surround Afghan population statistics at higher levels of analysis, provincial and national. In order to remedy this situation, three successive national surveys have been undertaken.

The first, in 1339 Š./1960, was a sampling carried out under the direction of the ministry of planning. From a list of 14,205 villages throughout the country compiled by the ministry of agriculture, a sample of 526 villages was chosen. Questionnaires were sent to the provincial governors, with instructions for their completion; they were actually administered, however, entirely by untrained local government employees, and most of the information was gathered from village headmen (maleks, arbāb). For the 413 villages that delivered usable information the total population was estimated to be twice that of registered males. An average population of villages was then calculated for each province and multiplied by the number of villages listed for the province; the result was an estimate of the total village population of the province. Statistical tests were applied to determine the sampling error in such calculations. As a result of this survey the sedentary rural population of Afghanistan was estimated at ca. 8 million, compared to the ministry of interior’s estimate of 10.4 million in the same year. Adding to this total a crude estimate of 1 million city dwellers and 2.4 million nomads (ministry of interior estimates, both obviously grossly inflated), a total population of ca. 11.4 million inhabitants was obtained for Afghanistan as a whole (Ministry of Planning, 1964).

As an extension of this first effort there was a second in 1346 Š./1967: a national agricultural census organized by the Ministry of Agriculture. As in the earlier instance, it was the village headmen who provided the basic data; the results must thus be viewed with caution. Curiously, the demographic findings of this survey were not included in the final publication, though they were published later in Atlas-e qarīahā-ye Afḡānestān (CO), along with the information from the civil registers. The disparities between these two sets of figures bear eloquent witness to the Afghan government’s total uncertainty about national demographic realities at that time.

The third national census was the only one to have had an exclusively demographic objective. The National Demographic Survey was launched on the initiative of the Ministry of Health and the Afghan Family-Guidance Association, with the cooperation of the U.S. Agency for International Development (AID), and carried out by a team from the State University of New York at Buffalo. It was a sampling survey that included 20,257 household interviews conducted in 1351-52 Š./1972-73: 9,070 (45 percent) in thirty-five urban centers and 11,187 (55 percent) in 352 villages distributed among eighty-five different districts in all the Afghan provinces. Methodologically this survey represented several innovations: First, rigorous principles of scientific sampling were observed for the first time in Afghanistan, thus guaranteeing that the sample would be representative; second, the census takers received thorough professional training, to ensure the reliability of the results; and, finally, in a special effort to include data on women, approximately 10,000 were interviewed about pregnancy histories in order to permit demographic estimates and provide data on family planning (Chu et al.). The results revealed a sedentary population of about 10 million, of whom 1.5 million were city dwellers. The median age of this population was eighteen years and life expectancy at birth 34.6 years. The natural rate of increase was 2.2 percent a year, which ensured a doubling of the population every thirty-one years; this rate was higher in the cities (2.5 percent) because of a lower mortality rate, which compensated for lower fertility (Table 23). The unusually high ratio of men to women (115.6) reflects both underreporting and a higher mortality rate among females. The average number of persons per household was found to be 6.23, higher in urban areas (6.60) than in rural ones (6.17). Geographical mobility was high: One quarter of the population lived in places different from those in which they had been born, 8.8 percent in different provinces. The proportion was higher among females, owing to matrimonial mobility (Table 23).

During the summer of 1353 Š./1974 the same team conducted a complementary survey among the nomad population. Unfortunately, none of the results from that part of the survey is available. The only information that has been issued is a preliminary estimate of about 1.2 million for the total nomad population, half the official estimate (Spitter and Franck, p. 11).

Despite their sophistication, these samplings have yielded only an approximate picture of the demographic realities. At the stage of development that Afghanistan had reached by 1352 Š./1973 more exact knowledge of its population could be gained only from a full-fledged national census. Such a census project was officially launched in 1354 Š./1975 with the financial and technical assistance of the U.N. Fund for Population Activities (UNFPA). Important preparatory work was required, however. It took more than two years to correct and update the fifteen-year-old topographic maps on a scale of 1:50,000, to number in red paint the main doors of all housing units in the country, and to make preliminary lists of all households, both settled and nomadic, in each locality, in order to determine the census requirements. This phase eventually led to the division of Afghanistan into 10,000 “enumeration areas,” each comprising an average population of ca. 250 households, or 1,500 persons, to be surveyed by a single census taker during the twenty-day scheduled enumeration phase of the census. Final adjustments in method, questionnaires, and tabulation plans were made after two pilot censuses of both settled and nomad populations in Mīzān 1356 Š./October 1977 and Saraṭān 1357 Š./July 1978. The main census, originally planned for Sonbola 1357 Š./September 1978, was delayed for nine months following the revolution in Ṯawr 1357 Š./April 1978; it finally took place between 25 Jawzā/15 June and 13 Saraṭān/4 July 1358 Š./1979.

The times were extremely unfavorable for such an undertaking. Antigovernment rebellions were raging in several parts of the country, which prevented the census takers from counting the entire population. Of 10,000 census takers—all teachers specially trained for the project—about eighty were killed by rebels. There were reported instances in which the filled-out questionnaires were destroyed before they reached Kabul; most notable was the destruction of the entire set of data sheets for Bādḡīs province. It is estimated that the census was almost complete in urban areas but only 40 percent complete for the settled rural population and 15 percent for the nomads. From a geographical point of view the coverage was good, even excellent, in the northern provinces and in those parts of the south inhabited by the Dorrānī tribes, which remained loyal to the authority of Kabul longer than their Ḡilzī neighbors to the east. On the other hand, it was mediocre in the northwest and downright bad in the center and the east (Figure 11). As reliable information on both settled and nomad populations had already been gathered during the preparatory phase of the census, however, reasonable estimates are available for populations of all enumerated or partially enumerated districts.

Only preliminary results of the census of the settled population have so far been published (CSO, 1360 Š./1981). They are based on a sample of 5 percent of the total questionnaires from the rural population and 10 percent of those from the urban population collected in all the enumerated districts. For districts that were not enumerated, data collected during the preparatory phase have been updated by calculating the growth rate extrapolated from the enumerated zones and adding on the estimated increases. Precise data on the geographical distribution of the Afghan population are thus now available for the first time, though they are not necessarily all equally reliable (Table 24, Figure 12).

Generally speaking, the results of the 1358 Š./1979 census confirmed the broad trends observed in the survey of 1351-52 Š./1972-73. The sedentary population had risen to 13,051,358, of whom 1,976,738 (15.1 percent) were urban dwellers. Life expectancy at birth was 40.7 years, one of the lowest in the world. The natural growth rate was 2.6 percent a year, which meant that the population would be doubled every twenty­-nine years. In contrast to earlier samples, in which both rural birth rate and urban mortality had been underestimated, it appeared that the natural increase was actually higher in the villages (2.7 percent) than in the towns (2.2 percent; Table 25). The sex ratio consistently showed a notable excess of men over women. The average number of people per household rose to 6.19 (6.29 in town, 6.16 in rural areas). The census also provided abundant data on the levels of education, marital status, and economic characteristics of the population. On the other hand, on the night before the census was to begin Prime Minister Nūr-Moḥammad Tarakī went on radio and television to ban the question on ethnic affiliation that had been included in the questionnaire, probably fearing that doubt would be cast on the official claim to preeminence for the Pashtun.

As far as the nomads are concerned, no findings from the census itself have ever been published. Furthermore, the Afghan authorities refused to grant official recognition to the results of the preparatory survey conducted in 1357 Š./1978, in which 140,760 households were counted, that is, 788,200 people, assuming 5.6 people in each nomad household (de Benoist, 1984, pp. 82ff.). Critical study of these findings shows that they reflect the reality fairly accurately: The magnitude of underestimation is on the order of 5 percent for the nomads and perhaps 25 percent for the seminomads. It is thus possible to suggest an absolute maximum population of approximately 1 million nomads and seminomads in 1357 Š./1978 (Balland, 1988, p. 176). The Afghan government continued, however, to claim a nomad population of 2.5 million, which supported the official estimate of the total population of Afghanistan at 15,551,000 in 1358 Š./1979 (CSO, 1359 Š./1980, p. 37), whereas the actual figure was probably only about 14 million. Since 1363 Š./1984 estimates of the nomad population have been cut sharply in official publications, to 1.5 million.

The 1360s Š./1980s in Afghanistan have brought demographic upheavals on such a scale that the 1358 Š./1979 census data are already out of date, even before they have been made public. Several surveys of Afghan refugees in Pakistan have been carried out, though with uncertain results (N. H. Dupree). A picture of the present geographical distribution of the Afghan population has, however, recently been constructed on the basis of the infrequent and somewhat tenuous available data (Grötzbach, pp. 79ff.). It seems that the situation is too unstable to permit analysis here.