Document not found! Please try again

Spatial pattern of fertility transition in Uttar Pradesh and ... - CiteSeerX

3 downloads 0 Views 4MB Size Report
ratio method suggested by William Brass (Brass and Coale, 1968) adjusts the level of observed age-specific fertility rates to the level of fertility indicated by.
GENUS, LXVIII (No. 2), 81-106

MALAY DAS* – SANJAY K. MOHANTY**

Spatial pattern of fertility transition in Uttar Pradesh and Bihar: a district level analysis 1.

BACKGROUND

Since independence, stabilizing the population has been a top priority in India’s development agenda. Empirical studies in India highlighted the societal benefits of limiting population growth (Coale and Hoover 1958; Dyson, 2004; Bhat, 2004; Datta and Mohanty, 2005). Studies from other developing countries also suggest that the slower population growth may indeed accelerate economic growth and advance overall economic well-being (Srinivasan, 1988; Kelley, 1988). Despite various efforts, the population of India in the last six decades has increased more than threefold, from 361 million in 1951 to 1210 million in 2011, an additional 849 million people (Registrar General of India, 2011a), largely due to natural increase. Though the country has achieved higher economic growth in recent decades (Bhattacharya and Sakthivel, 2004) and improved health and education (Planning Commission, 2011; Navaneetham and Dharmalingam, 2011), the overall state of human development remained low (Planning Commission, 2011). In the composite index of human development in 2011, India ranked 134th among 187 countries (UNDP, 2011). This argues for further efforts to reduce population growth so as to improve the state of human development in the country. In 1952, India was the first country in the world to launch a family planning program with the objective of reducing birth rates (MOHFW, 2000). During the first five decades of implementation, the family planning program in India underwent several changes, starting with a clinical approach (19511961), then subsequently moving to an extension education approach (19621969), Health department operated, Incentive based, Target-oriented, Timebound and Sterilization-focused program (HITS) approach (1969-1975), coercive approach (1976-1977), recoil and recovery phase (1977-1994) and the reproductive and child health approach (Srinivasan, 1998). All these approaches differ with respect to targets, choices of methods and implementation strategy. The reproductive and child health (RCH) approach (the current RCH approach), which was implemented following the 1994 Internation* International Institute for Population Sciences (IIPS), Mumbai, India. ** Department of Fertility Studies, International Institute for Population Sciences (IIPS), Mumbai, India. Corresponding author: Malay Das; e-mail: [email protected].

81

MALAY DAS – SANJAY K. MOHANTY

al Conference on Population and Development (ICPD), postulated that population policies should be viewed as an integral part of women’s reproductive health (Srinivasan, 1998). Additionally, the National Population Policy 2000 in India was introduced with its medium term objective of achieving replacement level fertility by 2010 and a long term objective of achieving population stabilization by 2045 (MOHFW, 2000). This, essentially, called for immediate needs of reducing fertility throughout the country. The fertility transition and population stabilization in India is of global significance as it is home to 17% of the world’s population (Planning Commission, 2008) with a low level of human development. The total fertility rate (TFR) in India had declined from 3.6 in 1991 to 2.6 in 2008; by 2008, half of the states had reached the replacement level of fertility (Registrar General of India, 2009a; 2009b). Though all states of India are experiencing fertility transition, the pace of change is not uniform across the states. The fertility level in two of the larger states, namely, Uttar Pradesh and Bihar remains high. The TFR of Bihar declined from 4.4 in 1991 to 3.9 in 2008 whereas in Uttar Pradesh, during the same period, it declined from 5.1 to 3.8 (Registrar General of India, 2009a; 2009b). These states are also lagging in many of the key socio-economic and health indicators and ranked at the bottom of the human development index (Planning Commission, 2011; Mohanty and Ram, 2011). There have been concerted efforts to increase the reproductive and child health services in these states, but they seem to have little effect on lowering fertility. Moreover there is large variations in the level of socio-economic development and population characteristics among the districts of Uttar Pradesh and Bihar (Planning Commission, 2011; Chaudhuri and Gupta, 2009; Ram and Mohanty, 2002; Ram et al., 2005). In this context, the paper examines the pattern of fertility transition in the districts of Uttar Pradesh and Bihar. It also explores the factors determining fertility variations across those districts. The states of Uttar Pradesh and Bihar were selected with following rationale. First, they are among the bigger states in India and lagging in many socioeconomic and demographic indicators. Secondly, the population stabilization of India is contingent on the future fertility scenario in these states as these states constitute one-fourth of India’s population. Of all the bigger states of India, the decadal growth rate of population during 2001-2011 was the highest in the state of Bihar (25.07%), followed by Uttar Pradesh (20.09%) (Registrar General of India, 2011a). Without a faster fertility decline in these two states, the replacement level fertility and population stabilization in India cannot be achieved (Planning Commission, 2008). Third, in the wake of decentralized planning, demographic indicators are often sought for policy and program implementation at the district level. Most studies in India have been carried out at the national and state level and thus do not provide estimates at the district level. 82

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

2.

DETERMINANTS OF FERTILITY CHANGE: SELECTED REVIEW

The factors contributing to fertility change has been extensively addressed in demographic literature. Davis and Black (1956) identified a set of 11 “intermediate variables” through which various socio-economic factors affect fertility, but these variables were difficult to measure. Bongaarts (1978) provided a comprehensive model and outlined four proximate determinants, namely; proportion married, use of contraception, induced abortion and postpartum infecundability, which explain large variations in fertility. There are a number of studies that examined the role of socio-economic factors on proximate determinants of fertility in India (Jain and Adlakha, 1982; Jain, 1985; Irudaya Rajan, 2005). These studies attributed the increase in age at marriage and use of modern contraception as the key proximate determinants of fertility change. In the 1970s and 1980s, the most significant factors factors of fertility change were female literacy, age at marriage, and reduction in childhood mortality (Jain and Adlakha, 1982; Dreze and Murthi, 2001; Murthi et al., 1995). These findings are also consistent with a study of India’s fertility decline between 1961 and 1991 (Arokiasamy, 1997). Studies outlined women’s education as a significant predictor of small family norms and fertility decline regardless of religion, culture and level of development (Vaidyanathan, 1988; Jejeebhoy, 1995; UN, 1995; Parasuraman et al., 1999; Dreze and Murthi, 2001). In recent years, the role of space, use of maternal health services and diffusion of contraception were added in explaining the variation in proximate determinants of fertility (Guilmoto, 2000; Mohanty and Ram, 2011). For instance, about two-fifths of the reduction in TFR in Chhattisgarh - a state whose level of socioeconomic development is similar to that of Bihar - was among the poor, while the TFR has not shown any change in Bihar (Mohanty and Ram, 2011). Evidence also suggests that fertility reduction in recent years in India is largely due to fertility reduction among uneducated and poor women (Bhat, 2002; McNay et al., 2003; Arokiasamy, 2009). There are few studies that attempted to explain the variation of fertility in the districts of India, mainly using data from the Indian census. Bhat (1996), using 1991 census data of selected districts of India, found that joint family (other than nuclear family), the proportion of Muslims, the proportion of scheduled tribes, child mortality, unmet need for contraception, and agricultural and child labor have strong positive effects, while female age at marriage, female literacy, media exposure, and population density and number of banks per 100,000 people have strong negative effects on fertility. Dreze and Murthi (2001), using district level data from the 1981 and 1991 census, showed that female education and child mortality are important factors in explaining fertility differentials in districts of India. While districts with a higher proportion of Muslims tend to have significantly higher fertility, it was not so with respect to scheduled tribes. Their study shows that region is an important factor in 83

MALAY DAS – SANJAY K. MOHANTY

explaining fertility differentials. A study based on 358 districts of India indicates that fertility at district level tends to vary by level of socioeconomic development and gender biases in kinship structure and it is significantly associated with child mortality and female labour force participation (Malhotra et al., 1995). Dommaraju (2012), using data from the National Family Health Surveys (1992-1993 and 2005-2006), showed that the decline in fertility in India over the last two decades was due to changes in marital fertility resulting from a longer birth interval among younger women. He also added that the age at marriage has a significant influence on fertility, particularly, in countries where child-bearing occurs predominantly within marriage. 3. DATA AND METHODS

In recent decades, there has been growing interest in providing the estimates of fertility in districts of India using indirect methods (Registrar General of India, 1989; 1997; Bhat, 1996; Prakasam et al., 2000; Guilmoto and Irudaya Rajan, 2002). The indirect methods that were used to provide district level estimates of fertility in India are: P/F ratio method (ratio of reported average parities (P) to average parity equivalent (F)) Arriaga Method, Rele Method, Bougue-Palmore’s method and the Reverse Survival (RSV) method. The P/F ratio method suggested by William Brass (Brass and Coale, 1968) adjusts the level of observed age-specific fertility rates to the level of fertility indicated by the average parities of women below 35 years of age. The reported children ever born (CEB) are transformed into the estimated age-specific fertility rate (ASFRs) and the current fertility rate is adjusted by using the P/F ratios. The Arriaga method (1983) is the modified Brass P/F ratio method when fertility is changing. The Registrar General of India, using the P/F ratio and Arriaga methods, provided district level estimates of fertility in India for 1981 and 1991, respectively (Registrar General of India, 1989; 1997). The estimated TFR showed considerable variation across districts in India. The Rele method and Bogue-Palmore’s method are regression-based methods used to estimate fertility. The Rele method (1967) assumes that for a given level of mortality, the Gross Reproduction Rate (GRR) is linearly related to the child-woman ratio (CWR) (ratio of children under 5 years to women of childbearing ages) and the birth rate has a curvilinear relationship with CWR. The Bogue-Palmore method is a regression technique of estimating fertility from a number of predictors such as the child-women ratio (ratio of 0-4 children to women aged 15-49), the infant mortality rate, mean age at marriage and the index of fertility age composition (Bogue and Palmore, 1964). Prakasam et al. (2000), using both the Rele and Bogue-Palmore methods, provided comparable estimates of fertility in districts of India for 1991. The RSV method is the most widely used indirect method of estimating fertility. In RSV method, the number of children enumer84

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

ated in a defined age (say 0-6 years) is reverse survived by the appropriate survival ratio to obtain the number of births in the last 6 years preceding the date of survey. The estimated births are divided by the estimated population to obtain the birth rate. Bhat (1996), using the RSV method, provided district level estimates of CBR and TFR based on population aged 0-6 years from 1981 and 1991 censuses. Guilmoto and Irudaya Rajan (2002) also used the RSV method on the 0-6 population from the 2001 census of India to estimates the CBR in districts of India. They used the ratio of TFR to CBR to derive the estimates of TFR. The age-specific fertility schedules from the National Family Health Survey of 1998-1999 (NFHS-2) of each state was used together with the age distribution of women of reproductive age (15-49) from the 1991 census to convert the estimated CBR to TFR. Beside these studies, Ram et al., (2005) using the regression method, estimated TFR in districts of India based on birth order statistics. They used the combined percentage of first and second order births from the District Level Household Survey of 2002-2004 (DLHS-2) to derive the estimates of TFR for districts in India. Since 1981, the census of India has also been collecting data on the number of births occurred during the one-year period prior to the date of enumeration. However, the fertility estimates based on the one-year births reported at the census are not consistent and tend to be underestimated (Registrar General of India, 2009c). More recently, Mohanty et al., (2012), based on the data on 0-6 population and by using RSV method, estimated the CBR and TFR for all the districts of India for 2001 and 2011. In this paper, the RSV method is used to estimate the CBR in the districts of Uttar Pradesh and Bihar. To understand the spatial pattern of fertility transition in Uttar Pradesh and Bihar, the TFR at district level is considered for three periods of time, namely, 1991, 1998 and 2008. The district level estimates of TFR for the period 1991 are borrowed from the census of India estimates (Registrar General of India, 1997). For 1998, we used Guilmoto and Irudaya Rajan’s (2002) estimates. The TFR of 2008 (TFR2008) is estimated from the estimated CBR of 2008 (CBR2008) using regression method. The data on the 0-6 population are collected from the census of India website (http://www.censusindia.gov.in). The methods of estimating CBR and TFR are described below. 3.1 Estimation of CBR

In order to estimate the CBR2008 using the RSV method, the total number of births in six years preceding the survey is obtained by dividing the population aged 0-6 years by the survival ratio from birth to age 6 years, i.e. the survival ratio of the 0-6 population.

85

MALAY DAS – SANJAY K. MOHANTY

Step 1: Computation of Survival ratio for bigger states of India In order to compute the survival ratio of the 0-6 population for each district in the respective states, it was first calculated for all bigger states of India (with a population equal or larger than 10 million) as: S0-6 = ((L0-1 + L1-4 + 2/5*(L5-9))/700000 where, S0-6 is the survival ratio of the 0-6 population, and L0-1, L1-4, and L5-9 are the life table populations in the age groups 0-1, 1-4, and 5-9 years of age, respectively, in the bigger states. The survival ratio of the 0-6 year old population for each bigger state is computed from state-specific life tables generated by using the age-specific death rates (ASDRs) of the respective states available in the SRS report of 2008 (Registrar General of India, 2009a). Since the life table population is calculated for five year age intervals (except in the first two age groups), the average of the life table population in the age group 5-9 is multiplied by 0.4 (2/5) to get the population in the 5-6 year age group. Here, it should be mentioned, that, in the SRS report of 2008, the ASDRs are available for the 20 bigger states of India. Thus, the survival ratio of the 0-6 population is computed for bigger states only. Step 2: Computation of Survival Ratio in districts of Uttar Pradesh and Bihar The state-specific survival ratios of the 0-6 population are regressed on the state-specific under-five mortality rates (U5MRs) of 2008. The state-specific U5MRs of 2008 are obtained from the SRS of India (Registrar General of India, 2009b). The survival ratios of the 0-6 population are regressed on the U5MRs available for the same states. The regression coefficients obtained at the state level are then used to obtain the district specific survival ratio. The regression equation used to obtain the survival ratio in the districts is as follows: SR0-6 = 1.00226 - 0.00121*U5MR where, SR0-6 is the survival ratio of the 0-6 population in each district. The coefficient values in actual decimal points have been used to calculate the survival ratio of the 0-6 population in the districts. However, for the current purpose, the coefficient values are rounded up to five decimal points. The U5MRs for the districts of the respective states are derived by using Brass2 indirect estima1

The state-specific abridged life tables are constructed based on age-specific death rates to obtain the survival ratio of the 0-6 population at the state level. The survival ratio for any specific age group of the population is derived from an abridged life table and is relatively accurate (Shryock et al., 1980). 2 The Brass indirect estimation technique is a technique for estimating infant and child mortality from survey data on the survival of children ever born. In this method, the probability of dying before attaining certain exact childhood ages is determined by using average children ever born and children surviving by age group of mother. William Brass was the first to propose this technique (Brass and Coale, 1968).

86

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

tion technique based on CEB and CS data by age group of mother. The data on CEB and CS for the districts are obtained from the District Level Household Survey-3 (i.e. DLHS-3), 2007-2008. The United Nations (UN) South Asian model life table (Shryock et al., 1980) is followed to estimate the U5MRs for the districts because it adequately represents the mortality patterns of countries in the South Asian region, including India. The UN South Asian model is most commonly accepted for providing the estimates of child mortality in India (Registrar General of India, 2009c). The child mortality patterns for women in the age groups 20-24 and 25-29 are used under the assumption that the births and deaths of children ever born, reported by women in these age groups, are reliable (UN, 1983). The estimated U5MR and survival ratio of the 0-6 population for the districts of Uttar Pradesh and Bihar are presented in Appendix 2 and Appendix 3, respectively. Step 3: Estimation of total births in six year prior to survey The number of births (B) in each district of the respective states is estimated as: B = (P0-6/SR0-6) where, P0-6 is the population in the 0-6 year age group in each district and SR0-6 is the survival ratio of the 0-6 population in each district. Step 4: Estimation of CBR Once the total number of births for each district is obtained, the district level CBR is estimated as: CBR = (B/7* P1 October, 2007)*1000 where, P1 October, 2007 refers to the district-level population in the respective states as of 1st October, 2007 which is the midpoint between March 2011 and March 2005. The mid-year population in the districts and respective states is computed using the annual exponential growth rate. 3.2 Derivation of TFR from CBR

The TFR2008 was derived by using the regression method (excluding intercept) based on state level time series data of CBR and TFR for the 1981-2008 period available from the SRS of India (Registrar General of India, 2009b). The TFR is regressed on CBR separately for both the states to derive the regression coefficients. The state specific regression coefficient is then applied to the estimated CBR to derive the TFR for each district in both states. In the case of Uttar Pradesh, the regression equation used to estimate the TFR is: TFR = 0.144*CBR 87

MALAY DAS – SANJAY K. MOHANTY

In the case of Bihar, the regression equation used to estimate the TFR is: TFR = 0.143*CBR Since the estimates of TFR are based on the births in the 6 years prior to census, the estimates correspond to the mid-period. It may be mentioned that the estimates derived from the 2011 census are referred to that of 2008. The estimates of TFR for the districts created during 1991-2001 and 2001-2011, are not available from published sources. Therefore, the estimates of TFR for newly created districts for the periods 1991 and 1998 are assumed to be the same as in the parent districts. The newly created districts during 1991-2001 and 2001-2011 in both Uttar Pradesh and Bihar are presented in Appendix 1. 3.3 Multilevel analysis

In order to determine the variation in fertility attributable to the differences between district level characteristics within the states, a multilevel linear regression (MLLR) analysis was carried out for the period 2008. In the present case, the analysis has been conducted using data of 109 districts in Uttar Pradesh and Bihar, where the TFR2008 is the dependent variable. The independent variables used in the analysis are: proportion of women marrying below 18 years of age as a proxy indicator of age at marriage, proportion of women using any modern contraceptive method, mean years of schooling for women, proportion of poor (i.e. level of poverty), under-five mortality rate, proportion of scheduled tribe population, proportion of Muslim population, and proportion of urban population (i.e. level of urbanization). The TFR and women’s years of schooling are transformed into a logarithmic form and under-five mortality is transformed into a logit scale (Bhat, 1996). The state dummy (1 for districts in Uttar Pradesh and 0 for districts in Bihar) has been used to determine the variance components at different levels (state level and district level). The estimates for all the variables (except proportion of urban population); proportion of women marrying below 18 years, proportion of women using any modern contraceptive method, mean years of schooling for women (individual level), proportion of poor, under-five mortality rate, proportion of scheduled tribe population and proportion of Muslim population, are derived from DLHS3. The proportion of poor is derived from a set of household economic proxies (household assets and amenities). Principal component analysis (PCA)3 is used 3

The principal component analysis (PCA) is a method of computing weights for individual indicators used to compute a composite index. In the recent past, some authors have used PCA for computing wealth index based on household assets and amenities (Gwatkin et al., 2000; Filmer and Pritchett, 2001).

88

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

in deriving a composite wealth index separately for rural and urban areas as health estimates differ significantly when separate wealth indices for rural and urban areas are used rather than a national wealth index (Mohanty, 2009). The state-specific poverty estimate provided by the Planning Commission of the Government of India (2007) is applied to the composite index in deriving the poor. The under-five mortality rates for the districts are estimated by using the Brass indirect method as explained in section 4.1. The proportion of the urban population is obtained from the 2011 census of India. 4.

RESULTS

4.1 Levels in CBR and TFR

The estimated CBR2008 for Uttar Pradesh, derived from RSV method, was 27 compared to 29.1 of SRS of 2008. Similarly, the estimated CBR2008 for Bihar derived from RSV method was 31.6 compared to 28.9 of SRS2008 (see Figure 1). Figure 1 – Comparison of estimated crude birth rate derived from reverse survival method and that of SRS in Uttar Pradesh and Bihar, 2008

On the other hand, the estimated TFR2008 for Uttar Pradesh derived from the RSV method was 3.9, close to that of SRS (3.8). However, the estimated TFR2008 for Bihar derived from RSV method was found to be higher (4.5) than the SRS (3.9) (see Figure 2). Our analysis shows that the pace of fertility transition during 1991-2008 was substantially lower in districts of Bihar 89

MALAY DAS – SANJAY K. MOHANTY

compared to that in districts of Uttar Pradesh. Moreover, some districts of Bihar (Madhepura, Saharsha and Supaul) have shown marginal increases in TFR, whereas all districts of Uttar Pradesh have experienced a reduction in fertility during the same period. The lower pace of the fertility transition in the districts of Bihar may be responsible for the relatively higher fertility in this state. The estimated CBR2008 and TFR2008 for the districts of Uttar Pradesh and Bihar are shown in Appendix 2 and Appendix 3, respectively. In Uttar Pradesh, the estimated CBR2008 was the highest in the district of Bahraich (37.7), followed by Balrampur and Siddharthnagar (34.1 in both), Budaun (33.1) and Sonbhadra (32.9), and the lowest in the district of Kanpur Nagar (18.4), preceded by Lucknow (20.9) and Jhansi (21.7). Similarly, in Bihar, the estimated CBR2008 was the highest in the district of Kishanganj (37.2), followed by Khagaria (37.1), Katihar and Purnia (36.6 in both), and the lowest in the district of Siwan (26.4), preceded by Patna (27.2), Saran (28) and Munger (28.3). While the majority of the districts in Uttar Pradesh (56 out of 71 districts) had a CBR between 18 and 30, the majority of districts in Bihar (26 out of the 38 districts) had a CBR of 30 and above. Figure 2 – Comparison of estimated total fertility rate derived from reverse survival method and that of SRS in Uttar Pradesh and Bihar, 2008

On the other hand, the estimated TFR2008 in Uttar Pradesh was the highest in the district of Bahraich (5.4), followed by Balrampur and Siddharthnagar (4.9 in both), Budaun (4.8) and Chitrakoot, Kheri, Shrawasti and Sonbhadra (4.7 in each), and the lowest in the district of Kanpur Nagar (2.6), preceded by Lucknow (3.0) and Jhansi (3.1). Moreover, out of 71 districts in 90

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

Uttar Pradesh, 42 districts had a TFR of between 3.0 and 3.9 and 28 districts had a TFR of 4 and above. None of the districts had reached the replacement level of fertility. Similarly, in Bihar, the TFR2008 was the highest in the district of Kishanganj and Khagaria (5.3 in both), followed by Araria, Katihar and Purnia (5.2 in each) and Madhepura (5.1), and the lowest in the district of Siwan (3.8), preceded by Patna (3.9) and Munger and Saran (4.0 in both). It should be noted that 36 of the 38 districts in Bihar had a TFR of 4 and above. The coefficient of variation in TFR for districts, indicative of variability in fertility levels, was 12.8 in Uttar Pradesh, compared to 9.2 for districts in Bihar. 4.2 Spatial pattern of fertility transition in Uttar Pradesh and Bihar

In order to understand the spatial pattern of fertility transition in Uttar Pradesh and Bihar, the district level estimates of TFR for the three periods - 1991, 1998 and 2008 - are plotted in Figure 3 (Uttar Pradesh), and Figure 4 (Bihar). The trends in the estimates of TFR in the districts of Uttar Pradesh and Bihar indicate that the transition in fertility was more apace in districts of Uttar Pradesh, while it was observed to be slow in districts of Bihar. However, the pace of transition in fertility was not uniform across the districts in these states. In Uttar Pradesh, the decline in TFR between 1991 and 2008 was the highest in the district of Kanpur Dehat (45.2%), followed by Firozabad (43.3%), Deoria (41.4%) and Bijnor (41.3%), and the lowest in the district of Bahraich (3.6%), preceded by Kheri (7.8%), Sitapur (9.8%) and Sonbhadra (11.3%). Moreover, between 1991 and 2008, out of 71 districts in Uttar Pradesh, 3 districts had experienced decline in TFR of less than 10%, 9 districts had experienced a decline in TFR of 10-20% and 59 districts had experienced a decline in TFR of more than 20%. On the other hand, 35 of the 38 districts in Bihar had experienced a reduction in TFR between 1991 and 2008, except the districts of Madhepura, Saharsa and Supaul (shown marginal increases in TFR during the same period). Also, the decline in TFR was the highest in the district of Munger (35.5%), followed by Siwan (29.6%) and Lakhisarai and Jamui (27.4% in both), and the lowest in the district of Araria (1.9%), preceded by Sitamarhi (2%) and Khagaria and Kishanganj (3.6% in both) during the same period. However, the level of TFR in the district of Sheohar remains stagnant at 5.0 during this period. Results also indicate that the pattern of fertility transition in districts of Bihar was not similar to that of Uttar Pradesh. In Bihar, half of the districts (19 out of 38) had experienced a reduction in TFR by more than 10% during 1991-2008, compared to more than 90% of the districts (68 out of 71) in Uttar Pradesh.

91

MALAY DAS – SANJAY K. MOHANTY

Figure 3 – Total fertility rate in districts of Uttar Pradesh

92

Figure 4 – Total fertility rate in districts of Bihar

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

4.3 Factors determining fertility differentials

The MLLR analysis is carried out to examine the factors determining fertility as well as to capture the variation in fertility attributable to differences in state level and district level. The results of MLLR analysis for all the districts of Uttar Pradesh and Bihar are presented in Table 1. In MLLR analysis, four different regression models are used and the dependent variable is in logarithmic form in all models. Different models are developed in order to capture the interaction effects, if any, between the variables. In Model 1, the proportions of the Muslim population, scheduled tribe population, urban population, poor and under-five mortality rate are included. In Model 1, the level of urbanization has a significant negative effect on the TFR, with a 10% increase in the level of urbanization leading to a 2% decline in TFR. The proportion of the Muslim population, proportion of scheduled tribe population, poverty and under-five mortality rate have significant positive effects on the TFR. The variance components (random effect parameters) indicate that 54% of the variation in TFR is attributable to the state level, while 45.6% of the variation in TFR is attributable to the district level. In Model 2, use of any modern contraception is added to the variables already used in Model 1. All six variables used in Model 2 are seen to have a significant effect on TFR. The effect of modern contraceptive use, though not strong, is found to be negative. In Model 3, the proportion of women marrying before 18 years of age is included along with the six variables used in Model 2. In Model 3, level of poverty and use of modern contraception are not significant. The proportion of women marrying before age 18 is significant and positively associated with TFR, indicating that an increase in the proportion of women marrying below age 18 may lead to an increase in fertility. Moreover, the inclusion of this variable in Model 3 has accounted for some of the variance in TFR and, as a result, the variance component corresponding to the random intercept decreases by 23% from Model 2 to Model 3 (from 0.00706 in Model 2 to 0.00544 in Model 3). The inter-class correlation from model 3 implies that about 50% of the variation in TFR occurs due to variation in the district level characteristics of the states under consideration. In Model 4, women’s years of schooling is added to all the covariates in Model 3. In this model, all variables except use of any modern contraception and level of poverty are significant. From this model, it is also observed that the value of the random intercept decreases further by an additional 27.4%, reflecting the fact that the inclusion of women’s years of schooling must have accounted for most of the variance in the dependent variable. Furthermore, the random parameters in Model 4 indicate that 54% of the variance in TFR is attributable to the district level differences, while 46% of the variance in TFR is attributable to the state level differences, controlling for all variables. 93

MALAY DAS – SANJAY K. MOHANTY

Table 1 – Results of multilevel linear regression based on the data of 109 districts in Uttar Pradesh and Bihar

@Used in logit form. #Used in logarithmic form. -- Not included in the model. ***Significant at 1% level. **Significant at 5% level. *Significant at 10% level. Notes: The dependent variable is TFR2008, which is in logarithmic form. Figures in parentheses corresponding to the coefficients of Fixed-effects parameters represent absolute values of z-statistics. Figures in parentheses corresponding to the random intercept and variance represent the amount of percentage variance.

This analysis demonstrates that factors such as women’s years of schooling, proportion of women marrying below age 18, proportion of Muslim population, proportion of scheduled tribe population, level of urbanization and under-five mortality rate have a significant effect on TFR, indicating that these variables are important predictors of fertility. Moreover, the differences in district level characteristics account for major variation in TFR; thus, these variables are significant predictors of fertility variations across the districts in the states under consideration. 94

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

5.

DISCUSSION AND CONCLUSION

The fertility transition and population stabilization in India is of global significance due to its size and regional diversity in the level of socio-economic development. Though fertility transition began in early 1970s, it is uncertain when India can achieve the replacement level of fertility. While half of the states of India have reached replacement level of fertility, the four larger states of India (Uttar Pradesh, Bihar, Madhya Pradesh and Rajasthan) continue to have high fertility. The population stabilization in India is largely contingent on the future fertility scenarios in these states. Moreover, though demographic research has extensively dealt with the factors governing fertility change at the micro level, there are a limited number of studies that examine the variation in fertility at the district level in India. There seems to be a great degree of variation in fertility levels among the districts within the two demographically-speaking largest states of India, Uttar Pradesh and Bihar. The primary aim of this study was to understand the fertility transition in the districts of Uttar Pradesh and Bihar in the last two decades. Taking advantage of the results of the 2011 census of India, this study estimated two indicators of fertility, namely CBR and TFR, with the utmost care. It should be noted that, although large-scale surveys bridged the data gap in many population and health parameters at the state level, the fertility and mortality estimates are still not available for districts of India. For example, the DLHS-3 did not collect information on birth history and the fertility estimates are not available in published reports. Some researchers provided the indirect estimates of TFR for districts of India using birth order statistics from the DLHS-2 (Ram et al., 2005). Comparing the districts level estimates of TFR based on birth order statistics from DLHS-2 with those from the 2001 census produced by Guilmoto and Irudaya Rajan (2002), it was found that the estimates derived from the DLHS-2 data were relatively higher. Alongside with the decentralized planning and limited resources, fertility estimates are often required for effective program intervention. Results indicate that fertility reduction in districts of Uttar Pradesh during 1991-2008 was at varying degree, whereas it was slow in most districts of Bihar. However, the fertility levels, as of 2008, remain higher in many districts in these states. In Uttar Pradesh, 42 of the 71 districts had a TFR in the range of 3.0-3.9 and 28 districts had a TFR of 4 or more. On the other hand, 36 of the 38 districts in Bihar had a TFR of 4 or more. Surprisingly, none of the districts in these states had reached the replacement level of fertility. Thus, it is clear that, despite the observed reduction in fertility, many districts in these states continue to have very high fertility. This is a worrying sign as the increased population nullifies the developmental efforts in these states. The high fertility may not only affect the average progress at the household and individual level but also affect the average progress at the macro level. The results of the MLLR analysis indicate that factors such as women’s years of schooling, age at marriage for women, proportion of Muslim population, propor95

MALAY DAS – SANJAY K. MOHANTY

tion of scheduled tribe population, level of urbanization and under-five mortality rate are significant predictors of fertility differentials across the districts of Uttar Pradesh and Bihar. Though many of the districts in these states have recorded significant increases in female literacy in last two decades, TFR did not decline as expected. Emphasis on the use of modern family planning methods, increase in age at marriage of girls and reducing early childhood mortality may help in reducing fertility in these states. From our analysis, it is evident that the effect of use of modern contraception on TFR is insignificant when controlling for age at marriage (Models 3 and 4 in Table 1). This is most likely due to its correlation with age at marriage of women or because of the aggregate analysis. However, past studies demonstrated that the increased use of modern contraception among women was one of the major causes of fertility decline in districts of India (Jain and Adlakha, 1982; Jain, 1985; Irudaya Rajan, 2005; Dreze and Murthi, 2001; Murthi et al., 1995; Arokiasamy, 1997). Based on our analysis, we suggest to reposition family planning and to increase the accessibility and availability of contraception in all districts in Uttar Pradesh and Bihar. There is a greater need for strong political commitment to make family planning successful. While both the states had seen substantial economic growth and increases in female literacy over the last two decades, there is a greater need to push for increasing the use of contraception and promoting later marriage among girls. Involvement of community leaders is essential to generate demand for contraception. The investment in family planning is likely to yield more progress at the state and country levels in the coming years. Though our study is an update on recent estimates and the determinants of fertility in districts of Uttar Pradesh and Bihar, it is required to put forward the reliability of our study. We have compared our state level estimates of CBR and TFR with those of SRS (Registrar General of India, 2009a) and district level estimates of CBR with that of Annual Health Survey (AHS) (Registrar General of India, 2011b) to understand the reliability in the estimates of CBR and TFR. We found that our state level estimates of fertility are close to the SRS estimates. The correlation coefficient of CBR derived from the RSV method and the CBR from the AHS is 0.75 for Uttar Pradesh and 0.78 for Bihar. This suggests that our estimates are fairly reliable. Despite these, we acknowledge some limitations in our study. First, we have used the provisional results of the 2011 census (the 0-6 population and total population) for the estimation of the CBR. Over the census years, the difference between the provisional and final population has been very small. For example, in the 2001 census, while the provisional population of 0-6 age group was 30.5 million in Uttar Pradesh and 16.2 million in Bihar, the final population of the same age group was 31.6 million in Uttar Pradesh and 16.8 million in Bihar. If the final estimates of the 0-6 population are significantly higher than that of the provisional estimates, our fertility estimates are likely to be lower. Second, potential biases may arise due to age misreporting in 2011 census. Researchers underscored the reliability of the proportion of the population in the 0-6 age group for estimating the fertility indicators preferring that age 96

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

group to the age groups 0-4 and 5-9 years (Bhat, 1996; Gulimoto and Irudaya Rajan, 2002). Researchers also acknowledged that age misrepresentation in India is rapidly decreasing due to increased literacy levels (Gulimoto and Irudaya Rajan, 2002). If the under-count of children aged 0-6 years still persists, it is likely to underestimate the fertility indicators. Third, we have used the Brass method to estimate U5MR for the districts of Uttar Pradesh and Bihar and linked it to the survival ratio of population aged 0-6 years. If mortality continues to be as high for the population aged 5-6 as that of the under 5-year age group, it is likely to bias the estimates upward. In other words, it may overestimate the number of births and birth rate. Though, theoretically, these biases are possible, we believe that such errors are minimal and that our estimates are reliable.

References ARRIAGA E. (1983), Estimating Fertility From Data on Children Ever Born, by

Age of Mother, International Research Document No.11, U.S Bureau of the Census, U.S. Government Printing Office, Washington, D.C., 20402. AROKIASAMY P. (2009), “Fertility Decline in India: Contributions by Uneducated Women Using Contraception”, Economic & Political Weekly, XLIV (30): 55-64. AROKIASAMY P. (1997), “Determinants of Demographic Changes in India: An Analysis Using New Transformation of Variables”, Demography India, 26 (1): 45-62. BHAT P.N.M. (1996), “Contours of fertility decline in India”, in SRINIVASAN K., Population Policy and Reproductive Health, Hisdustan Publishing Corporation (India), New Delhi. BHAT P.N.M. (2002), “Returning a Favor: Reciprocity between female education and fertility in India”, World Development, 30(10): 1791-1803. BHAT P.N.M. (2004), “Indian Demographic Scenario: Vision 2020” In Planning Commission (eds.), India Vision 2020: The Report, Academic Foundation, New Delhi. BHATTACHARYA B.B., SAKTHIVEL S. (2004), “Regional Growth and Disparity in India, Comparison of Pre- and Post-Reform Decades”, Economic & Political Weekly, 39(10): 1071-1077. BOGUE D.J., PALMORE J.A. (1964), “Some Empirical and Analytic Relations among Demographic Fertility Measures, with Regression Models for Fertility Estimation”, Demography, 1(1): 316-338. BONGAARTS J. (1978), “A Framework for Analyzing the Proximate Determinants of Fertility”, Population and Development Review, 4(1): 105-132. 97

MALAY DAS – SANJAY K. MOHANTY

BRASS W., COALE A.J. (1968), “Methods of Analysis and Estimation”, in BRASS W., DEMENY P., HEISEL D.F., LORIMER F., ROMANIUK A., WALLE E. VAN DE., The Demography of Tropical Africa, Princeton University

Press, Princeton, New Jersey. CHAUDHURI S., GUPTA N. (2009), “Levels of Living and Poverty Patterns: A District-Wise Analysis for India”, Economic & Political Weekly, XLIV (9): 94-110. COALE A.J., HOOVER E.M. (1958), Population Growth and Economic Development in Low-Income Countries, A Case Study of India’s Prospect, Princeton University Press, Princeton New Jersey. DATTA P., MOHANTY S.K. (2005), “India’s Future Population and its Socio-economic Implications by 2015”, Demography India, 34(2): 167-184. DAVIS K., BLACK J. (1956), “Social structure and fertility: An analytic framework”, Economic Development and Cultural Change, 4(4): 211-235. DOMMARAJU P. (2012), “Marriage and Fertility Dynamics in India”, AsiaPacific Population Journal, 26(2): 21-38. DREZE J., MURTHI M. (2001), “Fertility, Education and Development: Evidence from India”, Population and Development Review, 27(1): 33-63. DYSON T. (2004), India’s Population: The Future, in DYSON T., CASSEN R., VISARIA L., Twenty-First Century India: Population, Economy, Human Development, and the Environment, Oxford University Press., Oxford. FILMER D., PRITCHETT L. H. (2001), “Estimating wealth effects without expenditure data or tears: An application to educational enrolments in states of India”, Demography, 38 (1): 115-132. GWATKIN D.R., RUTSTEIN S., JOHNSON K., PANDEY R.P., WAGSTAFF A. (2000), Socio-economic differences in health, Nutrition and Population, HNP Thematic Group of the World Bank. GUILMOTO C.Z., IRUDAYA RAJAN S. (2002), “District Level Estimates of Fertility from India’s 2001 Census”, Economic and Political Weekly, February 16: 665-676. GUILMOTO C.Z. (2000), The Geography of Fertility in India, in GUILMOTO C.Z., VAGUET A., Essays on Population and Space in India, French Institute of Pondicherry. IRUDAYA RAJAN S. (2005), “Emerging Demographic Change in South India”, in GUILMOTO C.Z., IRUDAYA RAJAN S., Fertility Transition in South India. SAGE Publication, New Delhi. JAIN A.K., ADLAKHA A.L. (1982), “Preliminary Estimates of Fertility Decline in India during the 1970s”, Population and Development Review, 8(3): 589-606. JAIN A. K. (1985), “The Impact of Development and Population Policies on Fertility in India”, Studies in Family Planning, 16(4): 181-198. 98

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

JEJEEBHOY S.J. (1995), Women’s Education, Autonomy, and Reproductive Behav-

iour: Experience from Developing Countries, Clarendon Press, Oxford. KELLEY A.C. (1988), “Economic Consequences of Population Change in the Third World”, Journal of Economic Literature, (27): 1685-1728. MALHOTRA A., VENNEMAN R., KISHORE S. (1995), “Fertility Dimensions of Patriarchy and Development in India”, Population and Development Review, 21(2): 281-305. MOHANTY S.K. (2009), “Alternative wealth indices and health estimates in India”, Genus, LXV (2): 113-137. MOHANTY S.K., RAM F. (2011), “Spatial pattern of poverty reduction and fertility transition in India”, Population Review, 50(1): 62-78. MOHANTY S.K., ROY T.K., DAS M., DEVI N.S. (2012), Fertility Transition in Districts of India, 1991-2011, Paper presented in the Second Asian Population Association Conference, 26-29 August, Bangkok, Thailand. MINISTRY OF HEALTH AND FAMILY WELFARE (MOHFW) (2000), National Population Policy 2000, MOHFW, Government of India, New Delhi. MURTHI M., GUIO A.C., DREZE J. (1995), “Mortality, Fertility and Gender Bias in India: A District-Level Analysis”, Population and Development Review, 21(4): 745-782. McNAY K., AROKIASAMY P., CASSEN R H. (2003), “Why are uneducated women in India using contraception? A multilevel analysis”, Population Studies, 57(1): 21-40. NAVANEETHAM K., DHARMALINGAM A. (2011), “Demography and Development: Preliminary Interpretations of the 2011 Census”, Economic & Political Weekly, XLVI(16): 13-17. PARASURAMAN S., ROY T.K., RADHA DEVI D., PASWAN B., AROKIASAMY P., UNISA S. (1999), Role of Women’s Education in Shaping Fertility in

India: Evidence from NFHS, IIPS, Mumbai. (2000), Evaluation and Adjustment of 1991 Census Data and Estimates of Demographic Indicators: District Wise Analysis, IIPS, Mumbai. PLANNING COMMISSION (2007), Poverty Estimates for 2004-05, Government of India, Press Information Bureau, New Delhi, accessed online: http://planningcommission.nic.in/news/prmar07.pdf. PLANNING COMMISSION (2008), Report of the Working Group on Population Stabilization for the Eleventh Five Year Plan (2007-2012), Government of India, Planning Commission, New Delhi. PLANNING COMMISSION (2011), India Human Development Report 2011: Towards Social Inclusion, Institute of Applied Manpower Research, Government of India, Oxford, Oxford University Press. RAM F., MOHANTY S.K. (2002), Ranking of Districts in India for Area Specific Planning and Programme Interventions, International Institute for Population Sciences, Mumbai. 99 PRAKASAM C.P., MURTHY P.K., KRISHNAIAH S.

MALAY DAS – SANJAY K. MOHANTY

(2005), Human Development: Strengthening District Level Vital Statistics in India, International Institute for Population Sciences, Mumbai. REGISTRAR GENERAL OF INDIA (2011a), Provisional Population Total: Paper 1 of 2011, India Series 1, Census of India 2011, Office of the Registrar General and Census Commissioner, Government of India. REGISTRAR GENERAL OF INDIA (2009a), Sample Registration System Statistical Report 2008, Report No.1 of 2009, Government of India, New Delhi. REGISTRAR GENERAL OF INDIA (2009b), Compendium of India’s Fertility and Mortality Indicators 1971-2007 based on the Sample Registration System, Government of India, New Delhi. REGISTRAR GENERAL OF INDIA (2009c), District Level Estimates of Child Mortality in India Based on 2001 Census Data, Ministry of Home Affairs, Government of India, New Delhi. REGISTRAR GENERAL OF INDIA (1989), Fertility in India: An Analysis of 1981 Census Data, Occasional Paper No.13 of 1988, Registrar General, India, New Delhi. REGISTRAR GENERAL OF INDIA (1997), District Level Estimates of Fertility and Child Mortality for 1991 and their Interrelations with Other Variables, Occasional Paper No.1 of 1997, Registrar General, India, New Delhi. REGISTRAR GENERAL OF INDIA (2011b), Annual Health Survey Bulletins 2010-11, Office of the Registrar General and Census Commissioner, Government of India, available at: http://www.censusofindia.gov.in. SRINIVASAN K. (1998), “Population Policies and Programmes Since Independence (A Saga of Great Expectations and Poor Performance)”, Demography India, 27(1): 1-22. SRINIVASAN T.N. (1988), “Population Growth and Economic Development” Journal of Policy Modeling, 10(1): 7-28. SHRYOCK H.S., SIEGEL J.S., LARIMON E.A. (1980), The Methods and Materials of Demography, Vol. 2, Fourth Printing (rev.), U.S. Bureau of the Census, U.S. Government Printing Office, Washington D.C. UNITED NATIONS (UN) (1983), Manual X: Indirect Techniques for Demographic Estimation, Department of Economics and Social Affairs, Population Studies, No.81, United Nations, New York. UNITED NATIONS (UN) (1995), Women’s Education and Fertility Behaviour: Evidence from the Demographic and Health Surveys, United Nations, New York. UNITED NATIONS DEVELOPMENT PROGRAMME (UNDP) (2011), Human Development Report 2011, Sustainability and Equity: A Better Future for All, UNDP, New York. VAIDYANATHAN K. E. (1989), “Status of Women and Family Planning: The Indian Case”, Asia–Pacific Population Journal, 4(2): 3-18. RAM F., SHEKHAR C., MOHANTY S.K.

100

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

Appendix 1 Districts created during 1991-2001 and 2001-2011 in Uttar Pradesh and Bihar

101

MALAY DAS – SANJAY K. MOHANTY

Appendix 2 Estimated under-five mortality rate, survival ratio of 0-6 population, and estimated crude birth rate (CBR) and total fertility rate (TFR) in 2008 for districts of Uttar Pradesh

...Cont’d...

102

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

Appendix 2– Cont’d

...Cont’d...

103

MALAY DAS – SANJAY K. MOHANTY

Appendix 2– Cont’d

104

SPATIAL PATTERN OF FERTILITY TRANSITION IN UTTAR PRADESH AND BIHAR ...

Appendix 3 Estimated under-five mortality rate, survival ratio of 0-6 population, and estimated crude birth rate (CBR) and total fertility rate (TFR) in 2008 for districts of Bihar

...Cont’d...

105

MALAY DAS – SANJAY K. MOHANTY

Appendix 3– Cont’d

106