1.1 – Regional differences

9The Arab world does not offer a homogeneous picture ; on the contrary, it constitutes a patchwork of very different fertility levels. From Lebanon to North Yemen and the United Arab Emirates (national population), the total fertility rate (TFR) doubles (Table 2).

Figure 1

Map^* of total fertility rates in Arab countries, 1980-1985

Map^* of total fertility rates in Arab countries, 1980-1985

^* Thanks are expressed to Hervé Le Bras, the author of the computer cartography programmes used in Figures 1, 7, 9 and 12, for kindly making them available.

10The map of fertility in Arab countries (Figure 1) shows fairly clear-cut regional differences : fertility is higher in the Arab peninsula and the Levant, and lower in North Africa, while the countries overlapping Black Africa are in an intermediate position.

11Two countries stand out from their neighbours : Lebanon, with a lower fertility than other countries in the area, and Libya, with a higher one.

12Fertility started to decline in Lebanon well before the rest of the Arab world (Courbage and Fargues, 1975). The only national fertility survey in this country was conducted in 1971 (Lebanese Association for Family Planning, 1971). It yielded a TFR of 4.57 children per woman, while in the same year this index reached 6 children per woman in Egypt and Tunisia. A recent survey of the population of Beirut shows that fertility has continued its downward trend during the civil war (Zurayk et al., 1985). Consequently, the UNO estimate seems acceptable (Table 2).

Table 2

Total fertility rates (TFR) and existence of legal family planning organizations (FP) in the countries of the Arab League, 1980-1985 (in increasing order of fertility)

Countries TFR (children/woman) Existence of FP a) b) Yes No Lebanon 3.79 + Tunisia 4.8 4.92 + Egypt 5.23 + Morocco 5.7 6.44 + Algeria 6.1 6.97 + Somalia 6.09 + Sudan 6.58 + Kuwait * 6.5 6.15 – Iraq 6.66 + Qatar 6.76 – Democr. Yemen 6.87 + Mauritania 6.90 – Libya 7.17 – Syria 7.17 + Jordan 7.38 + Bahrain * 7.4 4.63 + Yemen Arab Rep. 7.9 6.76 + United Arab Emir. * 8.2 5.94 –

Total fertility rates (TFR) and existence of legal family planning organizations (FP) in the countries of the Arab League, 1980-1985 (in increasing order of fertility)

^* Female nationals only.

13For Libya, on the other hand, the estimates proposed by the UN are not based on any field survey. They are too unreliable for this country’s apparent anomaly to be discussed here [5].

1.2 – Variations in time

14The data are not sufficient for a proper comparison of the recent past in these countries. However, it seems probable that around 1960 total fertility was of the order of 7 to 8 children per woman in all of them (except Lebanon). The present differences therefore presumably reflect changes which have set in during the last three decades.

15Civil registration makes it possible to follow birth trends for five countries (Figure 2), and general fertility for four of them (Table 3) ; these cover almost the whole range of present fertility levels.

Figure 2

Recent birth trends in five Arab countries

Recent birth trends in five Arab countries

16In Tunisia and Egypt, the transition began early ; after Lebanon, they were the first two countries in the region in which fertility started to decline. They had been the pioneers of Arab family planning, the first programmes beginning in the 1960s. In a first stage, these programmes no doubt facilitated birth control for women who were already motivated to use it, rather than aroused new interest. This is what seems to have happened in Egypt. The quasi-stability of total fertility between 1937 (6.45) and 1960 (6.71) has been shown to be the result of two counterbalancing trends : a fertility rise for cohorts born before 1930, but a decline for subsequent cohorts, whose members started their reproductive lives towards 1950 (Fargues, 1986b). It was as though the official family planning programme adopted in 1964 met a demand which had been latent for ten or fifteen years.

17After following remarkably similar trends between 1960 and 1974, Egypt and Tunisia suddenly diverged. Whereas birth and fertility rates continued their regular downward trend in Tunisia, the birth (and presumably the fertility [6]) rate in Egypt started to rise after 1975. Several explanations may be put forward. The most plausible is perhaps that between the end of its fourth war with Israel (October 1973) and the signing of the Camp David treaty (September 1978), the Egyptian army was gradually demobilized. After years of separation, couples were reunited and made up on long-delayed births.

18The next in line is Algeria. With a birth rate initially higher than in the two countries discussed above, the first stage of fertility decline began ten or fifteen years later. Total fertility has dropped by three children over the last twenty years (Table 3), indicating that its rate of decline is now similar to that of Tunisia.

19In Kuwait and Syria, the fertility decline began even later, with an initial level of fertility exceeding that in Algeria [7].

2.1 – Between an old and a new equilibrium

20We have shown elsewhere (Fargues, 1987) how two institutions which combine to characterize Moslem marriage law, polygyny and repudiation, rely on two conditions not mentioned in the law : universal marriage and age differences between spouses at first marriage.

21Polygyny, that is, the right [8] for a man to contract another marriage while he is already married, and repudiation, that is, an easy means of divorce offered to the husband (generally in order to remarry), require an excess of females in the marriageable population. A simple diagram (Figure 3) shows how the association of two norms leads to an excess of single females on the marriage market. Leaving aside the differences in age structure between the sexes, the male and female population pyramids may then be superimposed, which is approximately the case at the ages when first marriages are generally contracted. If all females marry for the first time at age A₁ and all males at age A₂, the number of females who remain single, P₁, is higher than the corresponding number of males, P₂. The difference between these two populations represents the "surplus" S of young women who will leave their single status by marrying a man who is already married. This will either be a polygynous marriage, if he is married at the time, or a remarriage if he is divorced or widowed. The former is the more common case of remarriage, the combined effect of age differences between spouses at first marriage and excess male mortality meaning that men are rarely widowers at the ages when remarriage with a single female is possible. Out of 1,000 marriages, 14 are between widowers and spinsters in Algeria (1980) and 15 in Egypt (1977), compared with 30 and 34 respectively between divorced men and spinsters (Statistical Yearbooks of Algeria and Egypt).

Table 3

Changes in total fertility in countries for which data are available (children per woman)

Year Tunisia Algeria GTFR Egypt GTFR Kuwait * GTFR GTFR MTFR 1960 6.71 1966 6.89 1967 6.11 7.56 1968 6.14 7.39 1969 6.31 7.40 1970 5.99 6.85 8.36 7.21 1971 5.72 6.51 1972 6.74 1973 6.27 1974 5.99 1975 5.81 6.06 5.52 7.08 1976 5.33 5.90 1977 5.83 6.03 7.41 1978 5.30 5.47 7.27 1979 5.29 5.53 7.05 1980 5.18 5.53 6.95 6.60 1981 4.95 5.42 6.39 1982 5.11 6.37 1983 4.91 6.33 6.71 1984 5.02 6.09 6.60 1985 4.95 6.47 1986 4.95 5.41

Changes in total fertility in countries for which data are available (children per woman)

GTFR : sum of age-specific general fertility rates.
MTFR : sum of marital fertility rates estimated by number of births in a given year divided by a weighted mean of marriages in previous years, calculated for Tunisia because of the length of the series available.
^* Female nationals only.

Figure 3

The unbalanced marriage market

The unbalanced marriage market

22Comparison of the different Arab countries (Table 4) shows that an increase in the age difference between spouses tends to go hand in hand with an increase in either the polygyny or the divorce rate, or both. This can only be taken as a tendency, as the indices used – the only ones we were able to calculate – are not perfect.

23These indices do not make it possible to assess the roles played respectively by polygyny and divorce, the two regulators of a marriage market characterized by a surplus of females. This would require comparison of the relative frequency of polygynous and divorced men among first husbands. Indeed, it is not the same thing to measure polygyny as the frequency of the polygynous state among all married men (see Table 4) or as the frequency of polygynous marriages among all marriages (which would be necessary for comparison). In the four countries where the calculations were possible (Table 4), the second index was roughly double the first. This simply means that, in about one case out of two, polygynous marriage is followed by divorce from the first wife, making the man monogamous again. In Kuwait, for instance, 36 % of divorces ended a polygynous marriage (1985). The mean proportion of polygynous men among all married men, 5.6 % for the twelve countries presented in Table 4, could correspond to a frequency of polygynous marriages of the order of 10-12 %. Divorce therefore plays a major role in many Arab societies, polygyny a more modest one.

Table 4

Nuptiality indicators around 1980

Countries Age difference at first marriage between husband and wife * Polygyny Divorces per 100 marriages ** per 100 married men per 100 marriages Tunisia a) 4.1 years 0.5 0 14.8 Syria a) and b) 4.3 1.9 n.d. 7.0 Algeria a) 4.4 1.8 5.6 n.d. Morocco a) and b) 4.6 6.6 n.d. 25.7 Kuwait a) 4.8 7.0 11.9 29.5 Yemen b) 4.9 5.2 n.d. 25.3 Jordan a) and b) 5.2 3.8 n.d. 19.8 Lebanon a) 5.3 3.7 n.d. 8.0 Bahrain a) 5.3 5.4 11.1 26.8 Egypt a) and b) 5.9 3.8 6.5 20.8 Sudan b) 6.5 16.8 n.d. 17.4 Mauritania a) and b) 7.9 6.0 n.d. 45.3 Mean 5.3 5.6 – 21.9

Nuptiality indicators around 1980

^* Difference in SMAM (singulate mean age at marriage calculated from the proportions single by age at a given date).
^** This index is equal to either a) the ratio of divorces to marriages in a given year or b) the frequency of ever-married females aged 40-49 whose first marriage was dissolved by divorce.
The former underestimates the frequency of marriages that end in divorce, as the number of divorces is divided by a number of marriages larger than those actually at risk (interference of widowhood, increase in marriage cohorts).

24These four phenomena – universal marriage, age difference between spouses, polygyny, divorce – have varied through time, although managing to maintain a fragile balance. Certain Arab societies seem to have experienced a major change during the 20^th century : a plummeting divorce rate. In Algeria, for instance, it fell from 35 % of marriages at the beginning of the century to 10 % around 1960. Between 1945-49 and 1975-79, it dropped from 28 % to 20 % in Egypt and from 25 % to 16 % in Damascus, etc. (Fargues, 1987). This lower divorce rate implied that it had a lesser regulating effect on the marriage market. As the second regulator, polygyny, also decreased, the balance could only be maintained by a compensatory reduction in the mechanisms that produced a surplus of marriageable females.

25The first of these mechanisms apparently did not react to the drop in divorce and polygyny : no rise in the proportion of females never married has been observed [9]. It is the second mechanism which has adapted : age at first marriage has risen for women, thus steadily reducing the age difference between spouses during the last two or three decades. This later first marriage for females is confirmed to be general. It is observed in all the Arab countries in which adequate data exist for calculating it, and in Morocco, a detailed study showed that every province, urban as well as rural, was affected. Regional differences are, however, increasing : the highest level and the highest increase have been observed in Rabat, where mean age of women at first marriage rose from 21.2 years in 1971 to 25.2 years in 1982, while in the outlying provinces of Guelmin (urban) and Ouarzazate (rural), both the levels and increases were lowest (from 19.0 to 19.5 years and from 17.6 to 18.2 years respectively) (CERED, 1982).

26We are thus faced with a transitional situation, where a balance based on repudiation-remarriage and polygyny offset by a substantial age difference between spouses, gradually gives way to a new balance where equality in age at marriage becomes possible as inequality concerning divorce and remarriage disappears.

2.2 – Lower divorce rate and contradictory fertility trends

27In none of the Arab countries are there data which would permit the complex relations between fertility and polygyny to be analysed. It is consequently impossible to examine how a reduction in polygyny would affect fertility, and our study must focus on the other factor of change, divorce.

28An ambivalent relationship exists between divorce and childbearing. In Moslem Arab societies, divorce often ends an infertile union. Its traditionally high frequency can thus be regarded as a factor that favours high fertility : on the one hand, because it redistributes infertile, but not necessarily sterile, partners ; on the other, because the threat of repudiation may encourage a woman to have more children. From this point of view, a lower divorce rate would be expected to result in lower fertility. However, divorce also means that until a woman remarries she is not exposed to the risk of childbearing. From this point of view, a lower divorce rate would lead to higher fertility, by increasing the length of exposure to risk. In a first stage, the second effect seems to gain the upper hand. In Algeria, for instance, between 1906 and 1966, a significant correlation is found between a drop in the divorce rate and a rise in the birth rate (Figure 4). But these both occurred at the same time as other changes which could account for each of them separately, so that it is impossible to be certain to what extent the decline in divorce – that is, the stabilization of the marital family – has contributed to the fertility rise observed during the first part of the demographic transition in a variety of Arab countries.

Figure 4

Birth rate, divorce rate and proportion of females remaining single in Algeria, 1906 to 1986

Birth rate, divorce rate and proportion of females remaining single in Algeria, 1906 to 1986

29In a second stage, however, the relationship between divorce and fertility tends to be reversed. Simple arithmetic shows that when there is less divorce, the age of women at first marriage rises ; thus a drop in the divorce rate contributes to delaying reproductive life. The Algerian example shows that when ages of women at first marriage begin to rise noticeably – in this case, around 1970 – the birth rate starts to drop. This suggests a twofold change in the marriage pattern, towards a more stable union and towards greater age equality between spouses, which leads first to an increase, then to a decrease, in fertility.

2.3 – Later marriage and lower fertility

30Fertility has fallen in all Arab countries in which it can be measured, except for Yemen. Most of the decrease observed so far can be attributed to later age at first marriage.

31In only three countries are data published which enable the most recent trends in general (Table 5) and marital (Table 7) fertility to be studied by age of women. They represent three different examples of the demographic transition. Their data have been compared for three age groups, under-25, 25-34 and 35+, which correspond respectively to the beginning of married life, peak fertility and the end of childbearing.

32a) Under-25. General fertility has declined more than at other ages, whereas marital fertility has risen at all ages. This means that the lower fertility among young women can be attributed entirely to later first marriage. Indeed, the proportion of single women between the ages of 20 and 24 years has more than doubled during the last fifteen years in Tunisia and Kuwait, and more than quadrupled in Algeria. As for the higher marital fertility, this clearly reflects selection : early marriage is a custom which persists only in the most "traditional" social groups, those least inclined to limit their family size. Comparing general fertility and marriage data for Egypt, Coale and Shorter (1982) reached similar conclusions : a drop in early marriage played a major role in that country’s fertility decline.

33b) 25-34. Both general and marital fertility have remained practically stable in the countries which are in the earlier (Kuwait) and the later (Tunisia) stages of the demographic transition. Algeria, in which the initial level was higher than in Tunisia, maintains its higher fertility despite a decline which has continued in this age group.

34Thus once they are married, and whatever their age at marriage, women maintain until the age of 35 the fertility which prevailed in times of early marriage.

Table 5

Age-specific general fertility trends in Tunisia, Algeria and Kuwait (rates p. 1,000)

Year Age groups Total fertility rates 15-19 20-24 25-29 30-34 35-39 40-44 45-59 TUNISIA 1966 75 285 335 302 232 115 35 6.35 1970 43 261 299 269 202 95 29 5.99 1975 40 238 307 265 188 97 27 5.81 1980 36 207 296 250 170 77 23 5.29 1982 33 186 281 254 149 63 25 4.95 V – 2.1 – 4.7 – 1.7 – 1.4 – 4.5 – 2.3 – 0.5 – 0.089 ALGERIA 1970 114 338 388 355 282 152 42 8.36 1977 97 285 342 336 267 129 27 7.41 1984 27 224 322 293 227 104 20 6.09 V – 6.2 – 8.1 – 4.7 – 4.4 – 3.9 – 3.4 – 1.6 – 0.162 KUWAIT * 1970 121 339 375 315 225 59 16 7.21 1975 123 339 374 284 216 57 23 7.08 1980 89 291 367 263 225 59 25 6.58 1985 68 267 349 282 234 72 24 6.48 V – 3.9 – 5.3 – 1.7 – 2.4 + 0.7 + 0.8 + 0.5 – 0.054

Age-specific general fertility trends in Tunisia, Algeria and Kuwait (rates p. 1,000)

V : Mean annual variation : slope of the regression line calculated at each age on all available points (presented or not in the table).
^* Female nationals only.

35It has been found in various studies that patterns of family formation during the first years of marriage have not been greatly affected by the spread of contraception [10].

36In Tunisia, for instance, the mean interval between marriage and first birth does not vary with fertility determinants such as educational status and place of residence [11]. Whatever their level of lifetime fertility, women do not use contraception before the birth of their first child (Chekir, 1985). Contraception only affects births of fourth and higher orders (Baraket, 1985). Tunisian statistics permit marriage duration-specific fertility rates to be calculated for two periods with a four-year interval. During the first twelve years of marriage, the difference between the two periods is very slight : cumulated fertility at the twelfth wedding anniversary fell from 3.92 children per marriage in 1977-78 to 3.78 in 1981-82. After the twelfth anniversary, the difference becomes much more substantial : 2.30 children per marriage during the first against 1.79 children during the second period.

Table 6

Changes in the proportion of single females at ages 20-24 (p. 1,000) in Tunisia, Algeria and Kuwait

TUNISIA ALGERIA KUWAIT * 1966 270 1966 112 1965 161 1975 361 1977 310 1975 320 1984 590 1984 534 1985 391

Changes in the proportion of single females at ages 20-24 (p. 1,000) in Tunisia, Algeria and Kuwait

^* Nationals only.

Table 7

Age-specific marital fertility trends in Tunisia, Algeria and Kuwait (rates p. 1,000)

Year Age groups Total 20+ 15-19 20-24 25-29 30-34 35-39 40-44 45-59 TUNISIA 1966 405 401 378 327 252 131 43 7.66 1970 289 413 347 292 220 107 35 7.07 1975 385 447 369 289 205 108 32 7.25 1980 439 452 381 280 185 85 27 7.05 1982 450 435 373 288 162 70 29 6.78 V + 5.8 + 2.7 + 0.7 – 2.2 – 5.0 – 3.3 – 0.9 – 0.040 ALGERIA 1970 275 444 449 402 327 187 56 9.32 1977 431 439 410 374 300 153 36 8.56 1984 290 508 425 333 246 113 23 8.24 V + 1.1 + 4.6 – 1.7 – 4.9 – 5.8 – 5.3 – 2.4 – 0.077 KUWAIT * 1970 364 452 418 347 253 75 23 7.84 1975 438 516 444 319 242 68 30 8.09 1980 453 494 450 311 250 70 33 8.06 1985 384 456 438 326 269 86 31 8.03 V + 1.5 – 0.2 + 1.4 – 1.4 + 1.1 + 0.7 + 0.5 + 0.011

Age-specific marital fertility trends in Tunisia, Algeria and Kuwait (rates p. 1,000)

V : Mean annual variation (see Table 5).
^* Female nationals only.

37A closer study shows that differences exist at all marriage durations (Figure 5). Fertility during the first year of marriage was considerably higher during the second period. There may be two reasons that explain this change : a rise in pre-marital conceptions, and fewer marriages of younger girls, who are less fertile than their elders. On the other hand, between the third and the 25^th wedding anniversary, fertility was consistently lower during the second period, progressing with marriage duration [12].

38In Egypt, the pattern is very similar to that in Tunisia. Later first marriage means that births postponed from the years before the 25^th birthday are carried over into the next two age groups, 25-29 and 30-34, where general fertility rates have apparently remained relatively unchanged (Bucht and El-Badry, 1986).

39All these observations seem to suggest that contraception only intervenes once the family has reached its desired size, that is, in the next age group. In this case, is contraception as a means of birth spacing insignificant ? Are its effects merely limited, or even offset, by the fact that protracted breastfeeding, a well-known inhibitor of fertility, is gradually disappearing ? Examining the various proximate fertility determinants for two rural areas of Egypt, Beni Suef and Menoufia, Gadalla (1987) confirmed the results reported elsewhere (Page and Lesthaeghe, 1981). When women’s educational status improves, the gain due to more widespread contraception and the loss due to shorter breastfeeding practically cancel one another out. This leaves only the effect of age at marriage : the more education a woman has received, the later she marries.

40c) Over-35. Both general and marital fertility have fallen where the second stage of the demographic transition is already well under way (Tunisia and Algeria), but have tended to rise [13] in the country at the very beginning of this stage (Kuwait). The diffusion of contraception makes all the difference here, essentially contraception for stopping rather than spacing births.

Figure 5

Marriage duration-specific fertility in Tunisia, 1977-78 and 1981-82

Marriage duration-specific fertility in Tunisia, 1977-78 and 1981-82

Figure 6

Fertility by age at marriage after 15-19 years of unbroken marriage

Fertility by age at marriage after 15-19 years of unbroken marriage

41Age at marriage is a key variable for understanding differences in family size within a same country : lifetime fertility declines regularly as women marry later and later. However, it does not explain the differences observed between countries (Figure 6). This requires investigation into the specific determinants of fertility.

3.1 – The educational transition

43In developing countries, extension of schooling often goes hand in hand with a fertility decline. Several factors are involved :

44a) A purely demographic factor. Longer schooling contributes towards delaying first marriage, with the consequences we have discussed at length.

45b) An economic factor. Widespread schooling, in particular secondary schooling, is a sign of a transformation in the role of children. The balance between the gains from having children (in the short term, their labour, in the long term, "old age insurance") and the costs they entail (education, health, housing, food and clothing consumption), is reversed. Whereas in traditional rural society gains exceeded costs, the latter now tend to exceed the former, particularly in towns, where children start work later and the goods consumed by families are no longer produced at home.

46c) A psychological factor. Schooling projects girls into a world outside their family circle. It may give them occupational ambitions other than the traditional role of homemaker and mother.

47For all these reasons, a negative relation is expected between length of schooling and fertility level. This has, indeed, been observed within each of the Arab countries for which data are available (Table 8). However, this variable sheds little light on fertility differences between countries. An illiterate Tunisian woman gives birth to fewer children than a Jordanian woman with 4-6 years schooling. A Tunisian woman with 1-3 or 4-6 years schooling has fewer children than a Jordanian woman who has spent over 7 years at school, etc.

Table 8

Total fertility at the 25^th wedding anniversary by educational status (children per woman) : World Fertility Survey

Country Number of years the woman attended school 0 1-3 4-6 7 + Egypt 6.53 6.35 6.15 3.78 Jordan 9.73 9.27 7.70 6.19 Morocco 7.03 5.53 5.82 4.63 Syria 8.97 7.20 6.53 5.42 Tunisia 7.32 5.92 6.01 3.88

Total fertility at the 25^th wedding anniversary by educational status (children per woman) : World Fertility Survey

48We selected girls’ secondary schooling rates as an index of the three factors listed above. The map of national variations in this rate (Figure 7) does not correspond to the one representing total fertility (Figure 1). Among the countries with the highest schooling rates (shown in relief on map 7 bis), we find countries with moderate fertility (in light plane on map 7 bis), Lebanon or Egypt, and very high fertility (in dark plane), Jordan, Kuwait or Bahrain.

49The link between schooling and fertility thus disappears when international comparisons are made.

50Perhaps our indices were ill-chosen. The present rate of schooling is more closely linked to economic factors than to demographic and psychological ones. The schooling rate at the time when today’s mothers attended school would no doubt be more satisfactory for measuring these two factors. However, this cannot be done since no data are available.

51Another possibility is that schooling influences fertility in two ways. We have seen why it contributes to lowering fertility. There are perhaps also reasons working in the opposite direction. The extension of schooling was a factor in differentiating the sexes before becoming a factor of equality ; this may have confirmed the Arab woman in her role of wife and mother. If the balance between these contradictory effects varies from one country to another, the expected relationship between schooling and fertility will disappear.

52We proceed to examine how illiteracy has been reduced over the years. Tunisia is a striking example of what can be termed a real "educational transition" (Figure 8).

53Members of cohorts born at the turn of the century in rural areas, and a little earlier in urban ones, were in a situation of equal ignorance : the illiteracy ratio was high, and the probability of never attending school was the same for both sexes. As school attendance increased, so did inequality between the sexes. It is only in the urban cohorts born around 1950 and the rural ones born around 1960 that this gap started to shrink, and equality between boys and girls – this time in a situation of universal literacy – was not re-established until the cohorts born after 1970 in towns (and is yet to come in the country).

Figure 7

Map of secondary schooling rates in Arab countries, 1980-85

Map of secondary schooling rates in Arab countries, 1980-85

Figure 7

bis. – Total fertility rates (plane) and secondary schooling rates (relief) in Arab countries, 1980-85

bis. – Total fertility rates (plane) and secondary schooling rates (relief) in Arab countries, 1980-85

Figure 8

Illiteracy in Tunisia, 1984

Illiteracy in Tunisia, 1984

54The difference observed between illiteracy trends of males and females would be repeated, on a different scale, for all other levels of education. It can be taken to reflect mean educational inequalities between spouses. In Tunisia, these inequalities increased continuously throughout the first half of the 20^th century. 1950 marks a turning point, the gap having been steadily reduced for cohorts born after this date. The time when these individuals began their reproductive lives, around 1965, coincides more or less with the first stages of the fertility decline in Tunisia.

55In that country, the cohorts with the greatest educational inequality between spouses – the "turning-point" cohorts – are those born shortly before 1950. In other Arab countries, they will no doubt differ : schooling did not spread at the same pace everywhere, and the inequality between sexes did not increase, and then decrease, simultaneously in all countries. This may partly account for the strong divergence observed between the schooling and fertility maps.

3.2 – Towards a transition in family roles ?

56Childbearing and employment may be in competition with one another, particularly in the lives of urban women. They compete for time, which could be devoted to either child-rearing or paid work, and space, as the place of work is not necessarily the home. More generally, there is competition between two aspirations, family fulfilment and social achievement. This competition is experienced worldwide, but the problem may be particularly acute in Arab societies : it is there that urban women’s employment rates are the lowest in the world (Fargues, 1986a).

57The map of women’s participation in the labour force [14] (Figure 9) is very close to representing the converse of their fertility map. Except in the two most rural countries (Mauritania and Sudan, which are also more strongly influenced by Black African traditions, where women’s economic role is more readily accepted), fertility is lower (light plane on Figure 9 bis) whenever women’s employment is higher (in relief on Figure 9 bis).

58Does this result lead us to an economic interpretation of fertility patterns, which the dissimilarities observed between the international variations in schooling and fertility seemed to contradict ? Indeed, the practical necessity of leaving work as women have additional children should be regarded as one of the costs of childbearing. Confirmation of this mechanism would require cohort studies of the points during a woman’s life course when she enters and leaves the labour force.

59The economic assumption would imply that the employment rate falls as the number of children rises. As this rise goes hand in hand with an increase in the woman’s age, women’s participation in the labour force would be expected to decrease with increasing age. The last two population censuses in Tunisia confirm this (Figure 10 I). But as age increases, other changes occur, in the composition of the female population by marital status. The proportions of married women, and later of widows and divorcees, also grow, while the proportions of spinsters decrease. Thus lower employment at higher ages could perhaps be explained by marital status rather than family size. When age-specific rates of women’s employment are examined for each marital status (Figure 11 – for 1975 only, the 1984 census publications do not contain the necessary tables), it is this explanation which is confirmed. A clear-cut difference is observed between married women, 85 % of whom are unemployed whatever their age, and irrespective of whether they live in Tunis or elsewhere, and divorced or single women, between 30 and 40 % of whom participate in the labour force at the national level, and between 40 and 60 % in Tunis.

Figure 9

Map of women’s employment rates at ages 15 and over, around 1980

Map of women’s employment rates at ages 15 and over, around 1980

Figure 9 bis

Map of fertility (plane) and women’s employment rates (relief)

Map of fertility (plane) and women’s employment rates (relief)

Figure 10

Women’s employment rates in Tunisia : 1966, 1975 and 1984

Women’s employment rates in Tunisia : 1966, 1975 and 1984

60It is therefore marriage, not childbirth, which proves incompatible with women’s employment outside the home. The fact that many single or divorced women participate in the labour force shows that opposition does not stem from society, but from the conjugal family itself.

Figure 11

Women’s employment and marital status in Tunisia, 1975

Women’s employment and marital status in Tunisia, 1975

61How is women’s employment changing ? We have seen that it decreases with age, at a given date (1975 or 1984). Does this mean that it decreases as age advances, when we consider the same cohort through time ? Between 1975 and 1984, age-specific women’s employment rates increased substantially in Tunisia. Reclassifying the data by cohort (Figure 10 II) yields two superposed curves, which means that there was neither gain nor loss. In other words, the number of women who left work in order to marry was offset by the number of new arrivals on the labour market [15]. The new generations, those who reached their 15^th birthdays between 1975 and 1984, also participated in the labour force to a greater extent than their elders did at the same age. If their employment rate were not to decrease as they grow older, following the pattern of their elders, women’s role could, in the near future, be more profoundly modified than has been the case during recent years. The continuation of the demographic transition could well depend on the behaviour of these younger women as regards employment after marriage.

Figure 12

Map of rates of participation in the pilgrimage to Mecca, 1982-86

Map of rates of participation in the pilgrimage to Mecca, 1982-86

Figure 12 bis

Map of fertility (plane) and rates of pilgrimage to Mecca (relief)

Map of fertility (plane) and rates of pilgrimage to Mecca (relief)

62In Egypt, progress has been slower than in Tunisia. Between 1976 and 1987, women’s employment rates increased from 5.5 % to a mere 8.9 % [16]. This disparity between the two countries as regards women’s participation in the labour force can presumably be related to the disparity in their fertility trends.

63Does women’s employment have a direct influence on fertility, or simply represent a good indication of her position in the family and society ? Islam assigns a specific role to women, and the many regional variants of this religion no doubt contribute to the differences observed in their status. But this is a field where statistical data are extremely scarce : there are no figures of mosque attendance or surveys on religious practices. The Saudi Arabian Yearbook, however, publishes precious information on the number of pilgrims to Mecca by nationality. This provided us with a series of "pilgrimage rates" [17], whose international variations are remarkably close to those observed for fertility (Figure 12). However, this should not be taken to suggest that Islam constitutes a brake to fertility decline : the index is too rough for drawing conclusions of this kind. In addition to religious conviction, two other elements have considerable weight in the decision to make the pilgrimage : distance to Mecca [18] and standard of living. In this particular case, these two elements combine to favour certain countries, those nearest to Mecca also being those blessed with oil. For reasons which are not all related to the economy, it is these countries which are the most conservative as regards the family.