In Bangladesh, there has long been strong social pressure for the preservation of virginity until marriage, which is one of the cultural characteristics of the great majority of people in the country irrespective of their religion. Sex outside marriage occurs only seldomly, since pre-marital sex is looked down upon harshly in Bangladeshi society (Maloney and others, 1981). Marriage marks the beginning of the period of potential childbearing and, therefore, is considered the prime determinant of fertility in the face of the country's relatively low contraceptive use rate. Among females, almost 95 per cent of marriages take place before the end of their second decade of life. This densely populated country of 123 million people (ESCAP, 1996) is also characterized by a high population growth rate (2.2 per cent annually: BBS, 1994), high nuptiality and low age at marriage (Huq and Cleland, 1990; Islam and Islam, 1993). Nonetheless, there has been a clear rising trend towards higher age at marriage over time (Islam and Islam, 1993; Aziz and Maloney, 1985).

Traditionally, young age at marriage and early childbearing have been encouraged in Bangladesh. According to Aziz and Maloney (1985), Bangladeshi children, especially in rural areas, are socialized to assume their respective male and female roles well before puberty. This phenomenon has been observed more strictly among girls than boys, because of the impact of girls' behaviour during adolescence both on their own reputation and that of their family. Before the end of childhood, a girl is expected to begin learning proper decorum for a female so that she will be able to play the part well once puberty begins.

___________

* The authors of this article are M. Mazharul Islam, Associate Professor, Department of Statistics, University of Dhaka and Consultant, International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B), P.O. Box 128, Dhaka 1000, and Mamum Mahmud, Lecturer, Department of Statistics, Shahjalal University of Science and Technology, Sylhet-3100, Bangladesh.

The most dangerous stage of life of a Bangladeshi girl is the period following the onset of menstruation when a number of changes occur in her body, making her physically mature. At this stage, several restrictions are imposed by society on her movements, dress, food and freedom to make her own choices. When a young girl grows up, her parents keep her movements under surveillance. Such restrictions on the movement of unmarried girls and women sometimes serve to impede her education. She is advised at all costs to protect her virginity until marriage (Maloney and others, 1981).

In rural Bangladesh there are many social pressures to "marry off" pubescent girls (Aziz and Maloney, 1985). If the marriage of a pubescent girl is delayed, her parents, and sometimes the girl herself, are made to feel guilty. Sometimes neighbours and even relatives criticize parents if they have not married off their daughters soon after the onset of menarche. In such a situation, parents of poor socio-economic standing may begin to think of their daughter as a burden. Other factors also have an impact on this situation. In the Bangladeshi cultural context, younger females are in higher demand than older females as potential brides and they require less dowry as well.

Despite such pressures, there has been a growing concern in recent years that female children should not be married too young (Islam and others, 1995). Through a government order in 1976, the minimum legal age of marriage was fixed at 18 and 21 years for females and males, respectively. However, in the countryside, such requirements are hardly known and among those who do know about them, they have little impact on behaviour (Islam and others, 1995).

Frequently, experts have expressed concern about the negative social, health and economic consequences of teenage marriage, adolescent childbearing, unintended pregnancies and, in some societies, high levels of pre- and extra-marital conceptions (Bogue and others, 1977). Although such issues concerning adolescents are much publicized throughout the developed world (ESCAP, 1992), the concepts are relatively new in many developing countries such as Bangladesh.

This study examines the marriage patterns of adolescents and some socio-economic and behavioural characteristics of married adolescents in Bangladesh. It also analyses factors associated with adolescent marriage and draws out important policy implications from the findings of the study.

This study utilizes data extracted from the 1989 Bangladesh Fertility Survey (BFS) which was conducted during the period December 1988 to April 1989, on behalf of the Government of Bangladesh, by the National Institute for Population Research and Training (NIPORT), with funding from the World Bank. The details of the survey are available elsewhere (Huq and Cleland, 1990).

Although the 1989 BFS was not designed especially for studying adolescents, it did collect information through a nationally representative sample of 11,906 ever-married females under 50 years of age; this group comprised both adolescents and adults. Such a large data set provides an opportunity to study various aspects concerning adolescents, with the adults serving as a comparison group.

The sampling frame for the survey considered all households in Bangladesh from which a nationally representative sample of 11,729 households were selected, 11,236 of which were successfully interviewed. Among the 11,236 (7,984 rural and 3,252 urban) successfully enumerated households, a total of 12,096 ever-married females aged under 50 years were identified as eligible for individual interview. Of these, 11,906 females (8,467 rural and 3,439 urban) were successfully interviewed. Among the 11,906 ever-married females, 11,484 (96.4 per cent) had been married before age 20, of whom 1,922 (16.1 per cent) were currently under age 20. Univariate, bivariate and multivariate analytical techniques were used for data analysis.

Age patterns of marriage

Marriage among females in Bangladesh is virtually universal. Very few women remain single throughout their lives, as indicated in table 1 which shows the percentage distribution of single females in the population. The percentage who never married in the age group 45-49 years indicates the degree of permanent "celibacy" that exists in the society. Thus, the table shows that celibacy is virtually non-existent in Bangladesh. Most females have been married before age 20, with almost 100 per cent getting married by the time they reach age 30. However, table 1 also shows a remarkable change in the proportion of women single; it indicates a rising trend in age at marriage. Most of the change in the proportion never married has occurred below age 30; the shift occurred from 1974. The most remarkable rise in the proportion of never married females has occurred in the age groups 10-14 years and 15-19 years. The proportions never married at ages 10-14 and 15-19 rose from 73.4 per cent and 11.3 per cent in 1951 to 95.8 per cent and 48.6 per cent, respectively, in 1989.

Current age	1951 Census	1961 Census	1974 Census	1975 BFS	1981 Census	1989 BFS
<10	N.A.	N.A.	N.A.	99.8	-	99.6
10-14	73.7	67.4	90.5	91.8	98.0	95.8
15-19	11.3	8.3	24.5	29.8	31.3	48.6
20-24	3.0	1.3	3.2	4.6	5.1	11.8
25-29	1.1	0.5	0.9	1.0	1.3	2.2
30-34	0.5	0.4	0.6	0.2	1.0	0.2
35-39	0.2	0.2	0.4	0.4	0.4	0.1
40-44	0.2	0.2	0.5	0.2	0.7	0.2
45-49	0.2	0.2	0.3	0.0	0.3	0.1

Data from the 1989 BFS indicate that, among the 11,906 sampled married females aged below 50 years, about 96 per cent had been married when they were below age 20 and only 4 per cent were married at 20 years of age or older (table 2). Thus, it may be said that about 96 per cent of marriages in Bangladesh are teenage or adolescent marriages. This leads to an exceptionally low average age at first marriage, i.e. 14.8 years.

It is evident from table 2 that most teenage marriages take place at the ages 13 to 15 years; this age group accounts for 55 per cent of the total marriages. More than 80 per cent of teenage marriages occur at age 16 or younger. This pattern remains the same when the current age of married adolescents is controlled.

Married adolescents who were currently younger than age 30 at the time of the survey show higher mean and modal ages at first marriage than their older counterparts. The highest mean age at marriage (15.2 years) was observed among women who were aged 20-29 at the time of the survey. The corresponding figures for the age groups 30-34 and 40 and older are 14.3 and 14.0 years, respectively. Among the married women who were currently aged below 20 (i.e. adolescents) at the time of the survey, 8.6 per cent had been married by the age of 12. The proportion rises to 10.1 per cent for the cohort of married women who were aged 20-29 at the time of the survey. It rises further to 15.3 per cent and 23.3 per cent for the cohort of married women who were aged 30-39 and 40 and older, respectively, at the time of the survey. This indicates that the older cohorts had a lower mean age at marriage than their younger counterparts; it also confirms the trend towards increasing age at marriage in Bangladesh.

Current age	Age at first marriage												Mean age at marriage
	<9	10	11	12	13	14	15	16	17	18	19	20
< 20	0.1	0.8	1.4	6.1	20.0	24.0	20.3	15.1	7.9	3.0	1.2	-	14.5
29-29	0.2	0.9	1.4	6.9	17.1	17.5	16.9	13.9	9.1	6.7	3.8	5.5	15.2
30-39	0.5	1.6	3.2	9.4	20.8	17.6	15.7	12.0	7.3	5.2	2.8	3.9	14.3
40+	0.9	2.6	5.1	14.2	23.2	20.1	13.8	8.9	4.7	3.2	1.2	2.1	14.0
Total %	0.4	1.4	2.6	8.8	19.6	19.0	16.6	12.7	7.6	5.1	2.7	3.6	14.8
Cum %	0.4	1.8	4.4	13.2	32.8	51.8	68.4	81.1	88.8	93.8	96.4	100.0
N	47	162	306	1,043	2,336	2,267	1,975	1,511	911	604	319	423	11,906

Age at menarche and consummation

In the 1989 BFS, respondents were asked whether they had been married before or after their first menstrual period. In response, 18.1 per cent of the respondents who were married before age 20 reported that their marriage took place before their menarche, while 63.9 per cent reported that they were married after menarche and the remaining 18.0 per cent mentioned that their marriage and first menstruation occurred at the same time (table 3). Almost all the respondents who were married after age 20 reported that their marriage had taken place after the onset of menstruation. The results indicate that, on average, the girls attained menarche at the age of 13.4 years. The mean age at menarche is also in the neighbourhood of 13 years, which is in agreement with the finding of Rahman (1989). The results indicate that age at menarche strongly influences the timing of marriage in Bangladesh. There is a clustering of marriage soon after the onset of menstruation, as indicated by the fact that more than 70 per cent of the marriages took place soon after the onset of menstruation. The overall average age at consummated marriage was 15.1 years, which is nearly 3.5 months later than the average age at first marriage, i.e. 14.8 years. The 1975 BFS reported average age at consummated marriage as 13.1 years and the average age at first marriage as 12.7 years.

Marital stability

Table 3 presents some summary measures of marriage dissolution and prevalence of remarriage of married adolescents and adults. Among the ever-married women who had been married below age 20 (during adolescence), 13.9 per cent of the first marriages were dissolved: 6.2 per cent as a result of husband's death, 6.0 per cent because of divorce and 1.7 per cent because of separation. The corresponding figure for the ever-married women who had been married at age 20 or older (during adulthood) was 8.5 per cent: 2.5 per cent as a result of husband's death, 4.0 per cent because of divorce and 2.1 per cent because of separation. This shows that the percentage of widowed and divorced women is higher among those who got married as adolescents compared with those who married as adults. The percentage of women separated from their spouse is lower among those who got married as adolescents than those who married as adults. The results thus indicate that the overall marital dissolution rate is comparatively higher among those who married as adolescents than as adults.

Item	Age at first marriage
	Below 20	20 and older	All
Timing of marriage (%)
Before menarche	18.8	-	18.1
After menarche	62.6	100	63.9
Same time	18.6	-	18.0
Mean age at menarche (years)	13.4	-	-
Average consummation delay (months)	3.6	0.0	2.6
Mean age at consummation years)	14.8	21.2	15.1
Mean age at marriage (years)	14.5	21.2	14.8
First marriage status (%)
Married	86.1	91.5	86.3
Widowed	6.2	2.5	6.1
Divorced	6.0	4.0	5.9
Separated	1.7	2.1	1.7
No. of times married (%)
Once	93.6	97.9	93.8
Twice	5.9	2.1	5.7
Three times	0.5	-	0.5
Remarriage after dissolution of first marriage (%)
Yes	45.8	24.4	45.2
No	54.2	75.6	54.8
Average age (in years)at being:
Widowed	28.1	35.3	28.2
Divorced	18.4	23.5	18.5

Life-table analysis of marital dissolution also demonstrates the higher speed of first marriage breakdown among adolescents compared with adults. As many be noted from the life-table analysis of marital dissolution in table 4, marriages among adolescents were more unstable during the first few years of marriage than those among the married adults. Among the adolescents, close to 14 per cent of marriages had been disrupted even before the couple reached their first wedding anniversary (also by the end of the first year of marriage, this number increased slightly). The corresponding figure for married adults is only 8 per cent. In Muslim societies such as Bangladesh, women may remarry if their marriage is dissolved owing to the death of husband, or divorce. Table 3 shows that among the married adolescents whose first marriage was dissolved, nearly 46 per cent remarried compared with only 24.4 per cent of the married adults doing so. Table 3 also indicates that, of all ever-married women whose age at first marriage was below 20, 93.6 per cent were married only once; 6.4 per cent were married two or more times. The corresponding figures for married adults were 97.9 per cent and 2.1 per cent, respectively. The results thus indicate that the prevalence of remarriage is higher among married adolescents than their older counterparts. All the above findings, therefore, support the hypothesis that adolescent marriages are less stable than adult marriages.

Years since first marriage	Probability of remaining in first marriage
	Adolescents	Adults
0	0.86177	0.91841
1	0.85878	0.91183
2	0.85547	0.90670
3	0.85265	0.90489
4	0.84991	0.90994
5	0.84655	0.90526
6	0.84227	0.90909
7	0.84117	0.91556
8	0.83771	0.90500
9	0.83522	0.90960
10	0.83329	0.89809
11	0.82986	0.89394
12	0.82641	0.89474
13	0.82475	0.88119
14	0.82183	0.86207
15	0.81868	0.85915
16	0.81689	0.85484
17	0.81335	0.84000
18	0.80778	0.81395
19	0.80288	0.81579
20	0.79905	0.80645
21	0.79793	*
22	0.79509	*
23	0.79473	*
24	0.79285	*
25	0.78537	*
26	0.77966	*
27	0.77805	*
28	0.77601	*
29	0.77694	*
30	0.76660	*

* Frequencies less than 50.

Socio-economic differentials of adolescent marriage

Table 5 provides the percentage distribution of married adolescents (i.e. age at marriage less than 20) by selected socio-economic characteristics along with the same type of data for married adults (i.e. age at marriage greater than or equal to 20) for comparison purposes.

The results indicate that, among the total number of married adolescents, 92.7 per cent have a rural background, with the remaining 7.3 per cent coming from an urban area. The corresponding figures for the adults are more or less of the same (84.2 per cent coming from a rural area and 15.8 per cent from an urban area). When the childhood types of residence are controlled, the percentage of married adolescents who had been born and brought up in a rural area rises to 95.0 per cent; the urban childhood type of residence contributes only 5.0 per cent to adolescent marriages. The corresponding figures for married adults are 85.8 per cent and 14.2 per cent, respectively. It is evident, therefore, that teenage marriage is more prevalent in the rural rather than urban areas of Bangladesh.

As for region of residence, it may be observed that the frequency of adolescent marriage is highest in Dhaka division (30.5 per cent) followed by Rajshahi division (26.4 per cent) and Chittagong division (24.0 per cent), with the lowest incidence being in Khulna division (19.2 per cent). The same pattern follows for age at marriage being 20 years and older.

Among the married adolescents, about 70 per cent are illiterate or have no formal education; only 13 per cent have had seven or more years of schooling. Among the married adults, about 50 per cent have had no formal education, whereas 31.2 per cent have had higher education. This indicates that education may have a significant effect on teenage marriage, that is, education increases the age at marriage and consequently decreases the rate of adolescent marriage.

Almost 86 per cent of the married adolescents were women not working outside the home; only 14 per cent have had some experience working for pay. Among the married adults, 83 per cent have not had working experience, with the remaining 17 per cent having had some experience working outside the home. This clearly indicates that women's work status has very little effect on age at marriage. This is mainly because employment opportunities are very limited in Bangladesh and very few women get the opportunity to engage themselves in any kind of income-generating work before marriage.

Background characteristics	Age at marriage
	Below 20 (<20)	20 and older (< 20)
Place of residence
Rural	92.7	84.2
Urban	7.3	15.8
Region of residence
Chittagong	24.0	30.0
Dhaka	30.5	27.6
Khulna	19.1	14.5
Rajshihi	26.4	27.9
Childhood residence
Rural	95.0	85.8
Urban	5.0	14.2
Respondent's education
No school/Madrasha (religious school)	69.9	49.7
Lower primary	13.5	10.3
Upper primary	7.9	8.8
Higher	8.7	31.2
Work status (outside the home)
Yes	14.0	17.0
No	86.0	83.0
Husband's occupation
Professional/administrative/sales/services	31.0	49.8
Production workers	4.9	5.5
Non-agricultural workers	12.5	5.4
Cultivators/share-croppers	34.7	24.6
Agricultural labourers	13.8	10.8
Others, or not stated	3.0	4.0
Husband's education
No school/Madrasha (religious school)	52.5	34.0
Lower primary	13.3	10.6
Upper primary	8.0	6.4
Higher	26.2	49.0
Religion
Muslim	86.2	82.5
Non-Muslim	13.8	17.5
Economic condition (dwelling construction)
Poor	92.6	79.0
Middle	4.4	10.2
Upper	2.6	10.9
Ownership of agricultural land
Yes	57.0	59.1
No	43.0	40.9
Total	100.0	100.0
N	11,484	423

Among the married female adolescents, most of their husbands are illiterate, working as cultivators and day labourers. Only 26 per cent of the married adolescents reported that their husbands have had more than a primary-level education. The corresponding figure for the married adults is almost 50 per cent. About 70 per cent of the married adolescents' husbands were working as cultivators and day labourers as compared with to 50 per cent among the married adults. This indicates that adolescent marriage is prevalent mainly among illiterates. The data indicate that religion has a moderate effect on age at marriage. Hindus are less likely to be married below age 20 than their Muslim counterparts.

Ownership of land and the amount of land possessed have some differential effect on age at marriage. The percentage of marriages taking place below age 20 is higher (57 per cent) among women whose family has a few acres of cultivable land than their landless counterparts. As land possession is a mark of economic solvency, those having adequate land could easily fulfil demands for a dowry and other such expenses; thus, marriages in such families take place relatively earlier than in poor families.

Since data on household income and other indicators of economic condition seem to be unreliable, the status of dwelling construction was used as a proxy variable for economic condition by categorizing the respondents into three classes: poor (those whose dwelling is of ordinary construction), middle (those whose dwelling has walls or floors made with brick and roof made with metallic material) and upper (those whose dwelling is a modern concrete building). It has been observed that teenage marriages are most prevalent among the poor; 92.6 per cent of the respondents who had been married at ages below 20 were from the poor category described above. The corresponding figures for people assessed to be in the middle and upper classes were 4.4 per cent and 2.6 per cent, respectively. Thus, the prevalence of teenage marriage decreases in line with improvement in a person's economic condition. This observation also applies to age at marriage for adults.

Covariates of adolescent marriage

Logistic regression was used to identify the risk factor for adolescent marriage. In the model, age at first marriage was used as the dependent variable, which we dichotomised by assigning the value of 1 for age at marriage being less than 20 years and the value of 0 for age being 20 years and older; selected demographic and socio-economic characteristics were used as explanatory variables.

The explanatory variables considered in the model are as follows: place of residence, region of residence, childhood residence, respondent's education, husband's education, husband's occupation, work status of the respondent and availability of electricity in the household.

Table 6 gives the estimates of the logistic regression coefficients (B) corresponding to the independent variables, partial R and relative odds calculated for each category of the categorical variables. The category with relative odds of 1.00 represents the reference category for that variable. If the odds ratio is greater than unity, the probability of age at marriage occurring below 20 years is higher than that of age at marriage occurring at 20 years and older. P values are used to identify significant effects to assess the relative importance of the selected variables in the logistic regression.

From the results of the logistic regression analysis, it appears that respondent's education is the most important factor that influences age at marriage being below 20 years, when the other variables are controlled. Differential analysis also substantiates this finding. The relative odds of primary education (lower and upper primary) and no education are found to be 2.9 and 2.8, respectively. This shows that the likelihood of age at marriage being below 20 among women with primary education and no education is 2.9 and 2.8 times higher, respectively, than that of more highly educated women. Here it is also observed that prevalence of early marriage among women educated at the primary level is nearly the same as that of uneducated women, but the rate is very much higher than that of the more highly educated women. In part, this may be because the more highly educated women have a higher socio-economic status and live in urban areas. Thus, we conclude that age at marriage in Bangladesh could be raised by increasing the level of education for females.

Variables	Coefficient (B)	St.error of coefficient	Partial R	Odds ratio
Respondent's education	-	-	0.117	-
(Higher)	-	-	-	1.0
Primary	1.061**	0.161	0.108	2.9
No school	1.040**	0.165	0.103	2.8
Husband's occupation	-	-	0.043	-
(Labourers/farmers)	-	-	-	1.0
Land owners/cultivators professional/sales/services/production	0.124*	0.162	0.000	1.2
Workers	-0.310*	0.149	-0.025	0.7
Region of residence	-	-	0.043	-
(Chittagong)	-	-	-	1.0
Dhaka	0.277*	0.137	0.024	1.3
Khulna	0.489*	0.164	0.044	1.6
Rajshahi	0.008	0.138	0.000	1.1
Childhood residence
(Urban)	-	-	-	1.0
Rural	0.476**	0.162	0.043	1.6
Respondent's work status
(Yes)	-	-	-	1.0
No	0.346*	0.139	0.034	1.4
Husband's education	-	-	0.034	-
(Higher)	-	-	-	1.0
Primary	0.340*	0.164	0.025	1.4
No school	0.437*	0.157	0.040	1.5
Constant	2.438	0.274	-	-
Model Chi-square:	205.817
Degrees of freedom:	11
Probability:	.000

* Notes: Reference category is within parentheses; ** P<.01 and * P<.05.

The analysis shows that husband's occupation is the second most significant factor influencing low age at marriage. The logistic coefficients indicate that the highest occurrence of low age at marriage was among land owners/cultivators, followed by agricultural labourers or farmers, and skilled service employees. The odds against low age at marriage among the wives of land owners is 1.2 times higher than that of women whose husbands' are agricultural labourers and farmers. Another finding is that the relative chance of low age at marriage among the women whose husbands are engaged in sales and services, or in the professional sector, is 0.7 times lower than those of women whose husbands are agricultural labourers or farmers.

The analysis further shows that, with other covariates controlled, regional differentials in low age at marriage are also significant. The positive sign of each regression coefficient in table 6 suggests that, with reference to women who were living in Chittagong division, the residents of Dhaka, Khulna and Rajshahi divisions were more likely to have been married at a young age. The logistic regression analysis implies that the occurrence of low age at marriage among women who lived in Dhaka, Khulna and Rajshahi divisions is 1.3, 1.6 and 1.1 times higher, respectively, than that of women who live in Chittagong division. Thus, the occurrence of early marriage is almost the same in Chittagong and Rajshahi divisions while Khulna division shows the highest level of occurrence, followed by Dhaka division. This regional differentiation is observed because industrialization, urbanization and education did not evolve uniformly in all regions. For example, Chittagong division is the largest commercial and industrial area in the country. Dhaka division is relatively more urbanized and industrialized, having of a higher population density. Khulna division is comparatively less urbanized and industrialized, while Rajshahi division is rich in agricultural products and has more educational institutions than Khulna division. Demographic characteristics such as infant mortality and the sex-ratio of such deaths are also different for these regions (BFS, 1989).

As expected, the occurrence of low age at marriage is considerably higher in rural areas compared with urban ones. The analysis shows that rural women are almost 1.6 times more likely to be married earlier than urban women. There are some conceivable explanations for this difference. Females who migrated to urban areas were likely to have been married before they migrated. Most of them probably moved to the urban areas with their husbands who had employment there. The age at marriage for such women may be expected to be relatively low. In the case of married females who migrated to urban areas either with their parents or brothers before marriage, the age at marriage for them may also be expected to be lower, because their early socialization and their values related to family formation and family life may be somewhat different from females born and living in urban areas. Thus, migrant women might weigh down the actual age at marriage for urban areas (Ahmed, 1982).

The analysis indicates that the respondent's work status is the next most important determinant influencing low age at marriage. It has been observed that age at marriage below 20 years is likely to be 1.4 times higher among women who have never worked outside the home than those who had work experience. The primary cause for this situation may be the improvement in the socio-economic condition of the respondents as compared with others.

The analysis further shows that husband's education, though not as strong as the respondent's education, has a significant and positive effect on low age at marriage. The results (table 6) show that women whose husbands have been educated at the primary level had a significantly higher likelihood (odds of 1.4) of low age at marriage compared with those having a more highly educated husband. And women whose husbands are not educated were much more likely (odds of 1.5) to have been married at a young age than those with a more highly educated husband.

This study confirms that marriage is almost universal among females in Bangladesh; there are very few women who remain single throughout their lives. As in various other developing agricultural societies, early female marriage is customary in Bangladesh. Almost 100 per cent of females will have been married before the end of their second decade of life. Among the 11,906 ever-married women who constituted our study population, 96 per cent were married when they were teenagers, with most of such marriages taking place at the ages 13-16 years. Only 4 per cent of marriages occurred at ages 20 or older. The survey data indicate that a relatively large proportion (18 per cent) of the teenage marriages took place before the onset of menarche and there was a clustering of teenage marriage immediately after the menarcheal period. The mean age at menarche has been observed to be about 13 years. This situation gives rise to a very low age at marriage in Bangladesh, i.e. only 14.8 years, which is well below the minimum legal age for the marriage of females, i.e. 18 years, established by the Government in 1976. Legislation on age of marriage, therefore, seems to be ineffective in delaying childhood marriage in Bangladesh. In a recent study, Islam and others (1995) have observed that a large majority of the rural community in Bangladesh are ignorant about the legal age for marriage and are even less concerned about the negative social and health consequences of adolescent marriage.

It has been observed that older cohorts of women in Bangladesh have a lower mean age at marriage than their younger counterparts, which confirms that there is a trend towards increasing age at marriage, but the rise is at a very slow pace. During the last one and half decades, the average age at marriage has increased by only about two years, from 12.4 years in 1975 (BFS, 1975) to 14.8 years in 1989. In these circumstances, as the existing legislation is not working well and the prevailing cultural and social norms are unlikely to foster a delay in marriage, other avenues of possible policy intervention must be explored, such as the provision of increased opportunities for female education and employment outside the home for young women, both of which would be likely to delay marriage.

The results of this study indicate that the overall marital dissolution rate is comparatively higher among married adolescents than married adults. Thus, by raising the age at first marriage, greater marital stability could be achieved, provided no dramatic cultural transformation occurs in society.

Differential analysis shows that most of the married female adolescents have a rural background and have not had a formal education. Their husbands are also mostly illiterate and farmers by occupation. Logistic regression analysis identifies education, region of residence, place of residence (urban/rural), work status, husband's education and occupation as important covariates of teenage marriage.

The findings of our research hold implications for policy that could be useful in devising ways to solve the issues related to adolescent marriage and thus bring about a further reduction in fertility for Bangladesh. In order to enhance further the age at marriage and to reduce the rate of adolescent marriage, adolescents, their parents and the community should be made more aware of the negative health, social and economic consequences of early marriage, early pregnancy and large family size. This could be done through social mobilization, information, education and communication (IEC) campaigns, regular home visits by Family Welfare Visitors (FWVs) and Family Welfare Assistants (FWAs) (see article on pp. 15-26). In this context, the country's basic education system and its curricula should be redesigned to meet present day needs. Important would be education on family life, human sexuality, demographic, health, socio-cultural development, and the role of women in society, all of which are called for by the 1992 Bali Declaration on Population and Sustainable Development and the Programme of Action of the 1994 International Conference on Population and Development, both of which instruments Bangladesh is a signatory to. Education of girls as well as boys will not only ensure basic literacy for all, but also will provide a realistic basis for training women in income-generating activities and primary health care. The Government should also take appropriate measures to create more employment opportunities for young women and make efforts to employ more young women in white-collar jobs, especially in the fields of health and education. As early adolescent marriage seems to contribute to a greater likelihood of marital dissolution, a voice for adolescent women in making decisions about marriage should at least be promoted, if it cannot be ensured.

Ahmed, A.U. (1982), "Marriage and divorce in Bangladesh", Ph.D. thesis (unpublished), School of Social Science, University of Chicago, Chicago, United States.

Aziz, K.M.A. and C. Maloney (1985). Life Stage, Gender and Fertility in Bangladesh (Dhaka: International Centre for Diarrhoeal Disease Research, Bangladesh, ICDDR,B).

BBS (Bangladesh Bureau of Statistics) (1994). Bangladesh Population Census 1991, Analytical report, Vol. 1 (Dhaka: Bangladesh Bureau of Statistics, Ministry of Planning).

BFS (Bangladesh Fertility Survey) (1989). Second Analysis (Dhaka: National Institute of Population Research and Training).

Bogue, D.J., K. Oettinger, M. Thomson and P. Morse (1977). Adolescent Fertility: The Proceedings of the First International Conference on Adolescent Fertility (Chicago: Community and Family Study Center).

Chowdhury, A.K.M.A., S.L. Huffman and G.T. Curlin (1977). "Malnutrition, menarche and marriage in rural Bangladesh" Social Biology 24(4):316-325.

ESCAP (1992). "Adolescent Reproductive Behaviour: Asian and Pacific Region" Population Research Leads No. 41 (Bangkok: Economic and Social Commission for Asia and the Pacific).

__________ (1996). 1996 ESCAP Population Data Sheet (Bangkok: Economic and Social Commission for Asia and the Pacific).

Huq, M.N. and J. Cleland (1990). Bangladesh Fertility Survey, 1989, Main report (Dhaka: National Institute of Population Research and Training).

Islam, M.N. and M.M. Islam (1993). "Biological and behavioural determinants of fertility in Bangladesh: 1975-1989" Asia-Pacific Population Journal 8(1):3-18.

__________ and H.K.M. Yousuf (1995). Fertility and Reproductive Health Status of Married Adolescents in Rural Bangladesh (Dhaka: Population Development and Evaluation Unit (PDEU), Ministry of Planning).

Maloney, C., A.K.M. Aziz and P.C. Sarker (1981). Beliefs and Fertility in Bangladesh (Dhaka: International Centre for Diarrhoeal Disease Research, Bangladesh, ICDDR,B).

Rahman, M.M. (1986). "Age at consummation in Bangladesh", Chittagong University Studies (Science) 23:32-37.

For further information on this material please contact: loftus.unescap@un.org