Fertility convergence in the Indian states: an assessment of changes in averages and inequalities in fertility
ABSTRACT Are the fertility rates across the Indian states and socioeconomic spectrum converging? This paper seeks to answer to this question and makes three original contributions: First, a theoretical framework is conceptualised to understand the progress in fertility transition alongside socioeconomic and health transition. Second, the paper quantifies the progress in fertility transition across the major states of India. Third, fertility convergence is estimated both in terms of absolute and relative distribution of total fertility rates between the states and socioeconomic groups. The fertility transition plots and Changepoint estimates indicate varying pattern of fertility transition, critical Changepoints and steady positions among Indian states. The testing of regression based convergence measures for average fertility rates reveals that fertility rates were diverged over the longterm period of 19812009. However, the convergence measure estimates in subperiods suggest pronounced divergence of fertility rates during the initial period of 198191 but subsequently replaced with considerable volume of convergence for the recent period of 200109. Overall, India’s fertility transition is characterised by a transformation from progressive transition disequilibrium to progressive transition equilibrium in fertility rates for the Indian states and socioeconomic spectrum.
 Citations (2)
 Cited In (0)

Conference Paper: Is fertility converging across the Member States of the European Union?
Joint EurostatUNECE Work Session on Demographic Projections, Lisbon (Portugal), 2830 April 2010; 12/2010  Fertility differentials by religion in Kerala  a period parity progression ratio analysis. M Kulkarni . 21328.
Page 1
PERIANAYAGAM AROKIASAMY* – SRINIVAS GOLI*
Fertility convergence in the Indian states:
an assessment of changes in averages and inequalities
in fertility
1. INTRODUCTION
Long has been the idea of convergence, although, largely debated in the
economic literature, where it stems from the neoclassical model of economic
growth and it has a number of relevant policy implications (Lanzieri, 2010).
Though, the notion of convergence is also embedded in the theory of demo
graphic transition, until recently, it received comparatively little attention in
demography in examining empirical evidence for demographic convergence
(Oeppen, 1999). In the same vein, the classical demographic transition model
provides a scientific framework for studying many aspects of development
(Dyson, 2010). In recent times however, a growing interest in convergence
methodologies are found in demography (O’Connell, 1981; Watkins, 1990;
Herbertsson et al., 2000; Casterline, 2001; Wilson and Airey, 1999; Wilson,
2001; Coleman, 2002; Franklin, 2002; Reher, 2004; Moser et al., 2005;
Dorius, 2008, 2010; Lanzieri, 2010; Lee and Reher, 2011).
Focusing on demographic point of view, Watkins (1990) showed that,
during the 19thand 20thcenturies, there has been a tendency of greater demo
graphic homogeneity within nations. Wilson (2001) provided valuable quanti
tative assessment of the extent to which rising life expectancy and decline in
fertility have been converging, which led him to describe the process as ‘glob
al demographic convergence’(Dorius, 2008). Since the last decade, there were
growing evidences of convergence in fertility across globe and especially in
developed countries (Wilson, 2001, 2011; Reher, 2004; Franklin, 2002;
Dorius, 2008, 2010; Lee and Reher, 2011). Casterline (2001) modelled the
pace of fertility decline in less developed countries during 19502050 and
found a significant level of intercountry and interregional variation in the
pace of fertility decline. Dorius (2008) made more critical contributions to the
quantitative assessment of demographic convergence by estimating absolute
and relative fertility convergence rates. He incorporated population sizes and
inequalities in national fertility convergence estimates. Based on his results, he
challenged the notion that the last half of twenty first century was the period
of global fertility convergence. He stressed on the point that though, fertility
65
GENUS, LXVIII (No. 1), 6588
* International Institute for Population Sciences, Mumbai, India.
Corresponding author: Srinivas Goli; email: sirispeaks2u@gmail.com.
Page 2
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
66
rates across the globe converged over a longer period but with considerable
discrepancies. Moreover, reduction in fertility in world countries does not
mean that fertility rates are converging. He concluded that a fertility conver
gence phenomenon is a recent appearance on global scale (Dorius, 2010). On
the other hand, in a recent article, Wilson (2011) based on simple trend line
plots in fertility and life expectancy reiterated that the large majority of the
world’s population trends support the hypothesis of demographic convergence,
with no evidence of significant reversals with the exception of only few coun
tries still to embark upon it.
In sum, the notion of demographic convergence has generated consider
able curiosity around the question of assessing convergence by standard and
innovative methods in varying developmental context. However, most of the
earlier studies in terms of empirical attempts of quantification of the volume
of demographic convergence were focused on global context or between
groups of countries (e.g. Wilson, 2001; Dorius, 2008; Lanzieri, 2010). Very
few studies, for instance O’Connell (1981) Evans (1986) Alagarajan and
Kulkarni (1998) Bongaarts (2003) Franklin (2002), Alagarajan (2003), James
and Nair (2005) and Alagarajan and Kulkarni (2008) presented evidence that
countrylevel fertility rates conceal considerable differences in reproductive
behaviour among socioeconomic groups within countries. Further, quantitative
assessments of convergence models within and between developing countries
are rare. Within countries, fertility is usually higher in less developed region
(states) than developed region, higher in rural than urban areas, higher among
uneducated women than their bettereducated counterparts, and higher in
households with low incomes than their high income counterparts (Merrick,
2001). Asituation of this kind is ideal for testing convergence models to assess
the progress of within country convergence in fertility rates across the states
and socioeconomic groups.
Over the period, on an average, the progress in fertility decline in India is
remarkable (Rele, 1987; Registrar General of India, 19712007, 2009). The
total fertility rate for India fell from an average of 5.8 children per woman
(1951) to 2.6 children per woman (2009). India’s national demographic trends
are currently transitioning from third to fourth stage of demographic transition
(Visaria, 2004b). However, the progress in fertility transition is not uniform
across all the states and socioeconomic groups of India. Total Fertility Rates
(TFR) in India are declining with divergent destinies across states, ruralurban
and socioeconomic groups (Guilmoto and Rajan, 2001; Visaria, 2004b; James
and Nair, 2005; Kulkarni and Alagarajan, 2005; Alagarajan and Kulkarni,
2008). Comparatively, the south Indian states, urban areas, and higher socio
economic groups are approaching low fertility rates with some categories such
as the upper wealth and education quintiles of low fertility states reaching low
estlow fertility rate (TFR of 1.3), while many of the north Indian states, rural
areas, and disadvantageous socioeconomic groups still have higher fertility
rates (Registrar General of India, 19712007, 2009; International Institute for
Page 3
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
67
Population Sciences [IIPS] and Macro Internationals, 19922006). However,
there are some assumptions about current and near future fertility scenario,
which looks to be a period of continued ‘convergence’. Nevertheless, there has
been no formal attempt to estimate the volume and speed of fertility conver
gence across the Indian states and socioeconomic groups.
Being the second largest populous country with vast geographical varia
tions and faster pace of fertility transition, India fosters an ideal condition for
testing convergence hypothesis. Accordingly, the purpose of this paper is to
test fertility convergence hypothesis across sub national geographic units (the
states), and socioeconomic spectrum of India.
2. DATAAND METHODS
2.1 Data
This study utilized the following secondary sources of data: Sample Regis
tration System (SRS) and three rounds of National Family Health Survey
(NFHS) to assess fertility trends and the progress in fertility convergence across
the major states and socioeconomic groups (Registrar General of India, 1971
2007, 2009; IIPS and Macro International, 19922006). Since the early 1970s,
India’s Sample Registration System has been the most reliable source of fertili
ty estimates for the country and the states (Registrar General of India, 1971
2007, 2009). However, the SRS does not provide comprehensive trend data on
fertility by different social groups of India. Therefore, we have used the three
rounds of India’s National Family Health Survey to assess the trends and con
vergence in fertility across the states and socioeconomic spectrum of India.
NFHS is the equivalent to worldwide Demographic Health Surveys (DHS) in
terms of survey design and questionnaire. The NFHS surveys are widely used
source of information for estimating fertility and demographic indicators for
India and the states (for more information visit www.nfhsindia.org; IIPS and
Macro Internationals, 19922006). Additional data used in this paper include:
fertility rates for the historical period before 1971 from Rele (1987) estimates,
the population totals and literacy rates from the Census of India (Registrar Gen
eral of India and Census Commissioner, 19712011) and, poverty ratio estimates
from Planning Commission (Government of India, 19732006).
2.2 Methods
The methods of analyses of this study involve 1) defining and conceptual
ising the process of demographic convergence in general and fertility conver
gence in particular, and 2) adopting a range of measures and models to assess the
progress of fertility convergence in India.
Page 4
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
68
2.2.1 Defining and conceptualizing fertility convergence
In the economic literature, it is possible to find several definitions of con
vergence (see, for instance, Barro and SalaIMartin, 1991; SalaIMartin, 1996).
The first, most widely used concept describes the convergence of a group of geo
graphical units (states, regions, countries, etc.) as the reduction of their disper
sion of a given indicator such as TFR and IMR over time. In India, fertility rates
fell substantially in the recent decades but with considerable unevenness. As a
result, the Indian states and ruralurban areas are at different stages of fertility
transition. Closing of gap between the states, ruralurban and socioeconomic
strata over the time, more specifically fall in fertility levels to a same point can
be hypothesized as a process of fertility convergence. Therefore, the process of
the states and social groups becoming homogeneous in their fertility rates can be
termed as fertility convergence.
The pathways of the process of demographic convergence lie in demo
graphic transition. Moreover, the demographic transition process is also
described to go through the processes of equilibrium and disequilibrium via fer
tility and mortality convergence, divergence, and reconvergence across the geo
graphical units and socioeconomic spectrum. This study proposes that varying
timelines of the socioeconomic and demographic transition has segmented the
Indian states into different stages or regimes of demographic transition. We pro
pose a conceptual framework that illustrates the mechanism of interaction
between socioeconomic, health and fertility transition, resultant fertility conver
gence (see Figure 1).
Figure 1 – Conceptual framework: fertility transition and convergence model
Page 5
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
69
This framework envisages that during the progressive transition period, the
states and different socioeconomic groups improve at varying pace leading to
divergent trends until reappearance of post transition homogeneity or equilibri
um. The shifting process of progressive disequilibrium into equilibrium process
is termed convergence, and the time duration is called convergence period. Dur
ing the progressive equilibrium phase, the demographically weaker states expe
rience greater falls in fertility rates and come closer to states with lower fertility
rate, thus indicating the progress towards convergence process.
2.2.2 Measures of fertility convergence
Previous studies on convergence in fertility rates on global scale adopted a
variety of measures. However, a large number of studies used trend line plots and
classical economic convergence measures such absolute convergence, condi
tional bconvergence and Sigma convergence measures. In view of diverse stages
of fertility transition in India, in this study we have used both standard measures
of convergence (based on averages) and inequality based measures, which
account for relative and socioeconomic distribution of fertility with adjustment
for population sizes of the states. The methods adopted in this study have been
previously adopted for the study of convergence at global scale: between coun
tries convergence, convergence across European regions and subnational con
vergence in developed countries such as Italy and France (e.g. Herbertsson et al.,
2000; Wilson, 2001; Casterline, 2001; Coleman, 2002; Reher, 2004, 2007;
Moser et al., 2005; Dorius, 2008, 2010; Lanzieri, 2010; Lee and Reher, 2011).
In particular, the following comprehensive range of convergence metrics and
inequality measures have been adopted in this paper in an effort to assess the
progress of fertility transition, test convergence hypothesis and model the intri
cate patterns of fertility convergence among the crosssection of Indian states.
Absolute bConvergence Measure: absolute bconvergence measure is
used, where the gap between the rich and poor shrinks specially due to greater
pace of progress in lagging states and poorer social strata. For this measure, the
following linear regression model has been specified, following Rely and Mon
touri (1999):
Where is the mean annualized growth rate of the TFR in the
country i in the period (t, t+k), is the value in the initial time t and are
the corresponding residuals.
Page 6
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
70
Further, the speed of convergence is computed as: s = ln (1+tb) /t. Where
s = speed of convergence and tb‚ is the b convergence in t time period.
Conditional b convergence: When the analysis is focused at the national
level, it will not be reasonable to assume that all states will share the same
socioeconomic conditions; however, it is recognised that each state may be con
verging towards its own steady state across socioeconomic strata. This is referred
to as conditional b convergence and it may be detected with the inclusion of the
Barro regression of an additional set of variables that are likely to account for
varying socioeconomic conditions (Herbertsson, 2000). In this study, the condi
tional bconvergence measure is estimated by adding variables such as the per
centage of literates and the percentage of population below poverty line as
covariates in the absolute bconvergence measure. The equation of this model
can be written as
Where is the mean annualized growth rate of the TFR in the
country i in the period (t, t+k), is the value in the initial time t and
are the corresponding residuals. Similarly is the percentage of literates in
state i in the period (t, t+T) and is poverty ratio in state i and period (t, t+k).
Sigmaconvergence measure: Sigma convergence measure postulates that
convergence occurs when the dispersion of fertility rates decreases (Young et al.,
2008). Extending this logic to the case of Indian states, we expect the difference
between TFR across states to eventually shrink. The Sigma convergence model
was estimated as:
Where is the standard deviation (or assimilated measure) of the indica
tor at the time t. If the parameter reduces, it implies evidence of conver
gence.
Convergence Clubs: fertility may not converge strongly across all the
states but may converge more strongly among the states of a particular region
(MayerFoulkes, 2003). For example in India, the club of south Indian states
may be converging faster than other Indian states. In such a situation, the
incorporation of regional dummy in conditional bconvergence models is like
ly to indicate the picture of convergence within the club of south Indian states.
Inequality adjusted convergence measures, dispersion measure of fertil
ity: previous studies that examined inequality adjusted convergence adopted
various metrics such as Theil Index, Gini Coefficient, and Gaussian Density
function (Franklin, 2002; Dorius, 2008; Bloom, 2011). However, in view of
the widespread state and socioeconomic disparities in India, in this study we
Page 7
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
71
have used a modified and more robust test of convergence in fertility rates
based on the change in Dispersion Measure of Fertility (DMF) and Gini
Index, which account for ‘absolute’and ‘relative’inequality respectively. The
percentage reduction or increase in these two measures indicates either con
vergence or divergence in absolute and relative fertility rates, respectively.
The DMF model is a slightly modified version of Dispersion Measure of
Mortality (DMM) used by Shkolnikov et al. (2003) and Moser et al. (2004).
The DMF measures the degree of dispersion that exists at any point of time
in the fertility trend of states in a country. The DMF is calculated as the aver
age absolute interstate fertility difference, weighted by the size of population,
between each and every pair of states. Change in the DMF over time indicates
whether fertility rates are becoming more or less similar across the states; a
decrease in DMF indicates convergence, while an increase indicates diver
gence. The DMF for total fertility rate is measured in number of children per
women:
Where, i, j are states, and i, j
ity rate, W is the weighting, and
When applied to total fertility rate, TFR= Total Fertility Rate of state,
represents the relative population size of the state i.
Gini Index: lastly, to assess the relative inequality, the Gini Concentration
Index is used. For example in case of estimation of relative inequality in total fer
tility rate, G is equal to DMF divided by the average TFR
15, z is the India, TFR is the total fertil
3. RESULTS
The results of this study are presented in three sections: section one pres
ents results of the progress in fertility transition in India and the states, encom
passing the assessment of differentials in steady state: changepoint analyses.
Section two examines the closing of gaps in fertility rates across the states or the
catchingup process of higher fertility states with lower fertility states and polar
isation of cumulative population distribution to lower fertility rates. Section three
deals with a) quantification of the volume and speed of convergence in mean fer
tility rates (convergence in averages) across the major states and socioeconomic
groups and; b) quantification of the volume and speed of convergence in
absolute and relative distribution of fertility rates (convergence in inequalities)
across the states and socioeconomic spectrum.
Page 8
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
72
3.1 Fertility trends in India and states: changepoint analyses
The concept of demographic convergence lies at the heart of demographic
transition theory (Oeppen, 1999). In accordance with the conceptual framework
unveiled in figure1, the fertility transition process during the progressive stage
indicates divergent paths with some states experiencing faster pace of decline in
fertility rates e.g. Kerala, Tamil Nadu, Himachal Pradesh and Andhra Pradesh.
However, more proactive fertility control programme in the demographically
lagging states is likely to stimulate an acceleration of fertility decline leading to
a catchingup process with low fertility states.
Many previous studies that examined the demographic transition process
used trend line plots of fertility rates; however, simple trend line plots do not pro
vide critical changepoints in the transition process and the volume of change in
fertility rates. Therefore, in this paper, we have adopted a more recently
embraced innovative method of analysis namely ‘changepoint analyses’(Taylo,
2011). The assessment of variations in fertility transition in terms of critical
changepoints across the states helps in understanding the differentials in steady
state conditions across the states. The advantages of changepoint analyses are:
first, the changepoint analyses statistically estimate the critical changepoints
and the volume of change in fertility. Second, the changepoints more effective
ly characterize the changes in fertility with statistical significance or confidence
intervals. Third, it is a powerful tool to determine the changes in fertility more
accurately and controls the overall error rate is robust to outliers. Lastly, the
changepoint analyses are capable of detecting subtle changes in fertility missed
by simple trend plots (Taylor, 2011).
Table 1 provides the results of changepoint analyses for TFR in India and
its major states. The transition plots and table of significant changes for India as
whole indicate that over the more than half century period of 58 years India
experienced critical changes in fertility at five points: 1976, 1985, 1991, 1997
and 2004. The largest change in TFR (from 5.4 to 4.5) was observed in 1976 and
the magnitude of change was statistically significant at perfect 100% level of
confidence. In 1991, India experienced a greater change in TFR (from 4.50 to
3.51) at 99% confidence level.
The transition plots and table of significant changes by states show varying
patterns across the states. However, majorly three types of patterns have been
noticed: the first group comprises the south Indian states of Kerala, Andhra
Pradesh, and Tamil Nadu which experienced critical change in fertility transition
in late 1970s or early 1980s. The second group of states is Punjab, Karnataka, West
Bengal, Haryana, Assam, and Orissa which experienced critical change in fertili
ty transition from early 1980s to late 1990s. And, the last group is the states of
Madhya Pradesh, Rajasthan, Uttar Pradesh and Bihar that experienced critical
change in fertility in the post 1990. The critical changepoint estimates of fertility
for the states are statistically significant with greater confidence level (Table 1).
Page 9
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
73
Table 1 – Estimates of changepoint analyses: significant changepoints
for TFR of India and selected major states, 19502009
...Cont’d...
Page 10
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
74
3.2 Closing and catchingup process of gaps in fertility rates across the states
The estimates of critical changepoints across the states indicate varying
patterns of steady state in fertility transition for different states. The progress of
fertility decline in some states was faster during the 1970s and 1980s; however,
in the post 1990s phase, many of the demographically lagging states of India
were catchup with the demographically advanced states. This catchingup
process resulting in fertility homogenisation process across the states and socioe
conomic spectrum is more apparent from the results presented in this section.
The catchingup process has been examined with three types of measures: 1)
simple trend line plots of average TFR of states; 2) trend plots of the association
between change in TFR of entire period under observation on initial level of TFR
among major states of India; 3) and cumulative distribution of population of the
major Indian states by their total fertility rates during 19812009.
Figure 2 presents the simple trend line plots of fertility rates for the major
states of India. The figure shows that until 1991, the fertility rates across the major
states were diverged. This fertility divergence phase mainly occurred due to
greater relative decline in fertility in the demographically advanced south Indian
states compared with their north Indian counterparts. However, in the post 1990s
period the demographically advanced states experienced stabilisation in fertility
rates between TFR of 2 to 1.8; on other hand, the demographically lagging states
experienced greater pace of decline in TFR, thus, closing the overall gap in aver
age TFR levels across the states. Figure 3 presents complementing results for the
Table 1 – Cont’d
Page 11
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
75
catching up process. The relative change in TFR during entire period under obser
vation compared to initial (1981) TFR indicate that those states which have
greater initial TFR level (e.g. more than 4 and 5) have experienced greater change
in TFR visàvis those states which have lower levels of initial TFR have expe
rienced lesser change in TFR during 19812009. Such a pattern demonstrates
evidence of catchingup process in TFR across the 15 major states of India.
Furthermore, for greater empirical authentication of the catchingup
process, Figure 4 presents the cumulative distribution of population of the major
states of India by their total fertility levels during 19812009. The results indi
cate that over the period, a greater proportion of population is cumulating at
lower fertility levels. The pace of fertility change was more significant in the post
1990s period as cumulative population curve line of 2001 and 2009 are observed
greatly inclined to lower levels of fertility. The three types of TFR trend line plots
presented in this section indicate an emerging convergence phase in fertility rates
across the Indian states and thus portray an ideal condition for testing the con
vergence hypothesis. However, there is a need to quantify the volume and speed
of convergence for better understanding of the progress and priority setting in
policy interventions. Therefore, in the following section, we present results of
the volume and speed of convergence of TFR across the Indian states and
socioeconomic strata.
Figure 2 – Trends in mean TFR in major Indian states, 19812009
Page 12
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
76
Figure 3 – Change in TFR during 19812009 timeperiod in major
Indian states by TFR starting levels in 1981
Figure 4 – Cumulative distribution of the population of major Indian states
by TFR levels, 19812009
Page 13
3.3 Convergence estimates
The process of convergence in fertility rates across the states and socioe
conomic spectrum were examined with two categories of measures: 1) Con
vergence in average fertility rates was examined using absolute and condi
tional b convergence and Sigma convergence and; 2) Inequality based conver
gence in fertility rates was assessed using absolute and relative inequality
based convergence models.
3.3.1 Convergence in average fertility rates: Absolute‚ convergence model
estimates
In this section, we present estimates of absolute b convergence for TFR
of major states during 19812009. The results in Table 2 indicate that overall
fertility rates diverged across the states during 19812009. However, time dis
aggregated piecewise convergence model estimates for small intervals indi
cate evidence of fertility divergence only during the earlier period of 1981
1991. In contrast estimates for the recent periods of 19912001 and 2001
2009 show evidence of convergence in fertility rates across the states (b=
–0.01157, p<0.969 and b= –0.04577, p<0.834, respectively). The volume and
speed (S) of convergence was greater for the most recent period of 20012009
(S=5% per annum) compared with 19912001 (S=1% per annum).
3.3.2 Conditional b convergence model estimates
When estimating convergence rates for the states to the national level, it is
not correct to assume that all the states share the same socioeconomic steady
state. Therefore to account for and disentangle the effects of socioeconomic fac
tors on fertility rates, we estimated conditional b convergence model by incorpo
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
77
Table 2 – Absolute b convergence for Total Fertility Rate (TFR) among the
major Indian states, 19812009
Page 14
FERTILITY CONVERGENCE IN THE INDIAN STATES: AN ASSESSMENT OF ...
78
rating control variables. Conditional b convergence gives the convergence
rate, if fertility would not have been influenced by other factors. In this study,
we have estimated conditional b convergence by adding two critical socioe
conomic covariates namely a) the percentage of literate population and b) the
poverty ratio to Barro regression model.
Table 3 presents the results of conditional b convergence estimates. Sim
ilar to absolute b convergence estimates, conditional b convergence estimates
also show evidence of divergence in fertility rates across states in India dur
ing 19812009. Unlike, piecewise convergence estimates of absolute b con
vergence model that showed divergence in initial period and convergence in
recent period, the piecewise convergence estimates of conditional b conver
gence model show evidence of fertility divergence for all three periods (1981
1991, 19912001, and 20012009). However, the rate of divergence declined
for the recent period (i.e. 14% in 19811991 declined to just 1% in 2001
2009). Overall, the conditional b convergence model estimates indicate that
evidence of convergence in terms of absolute b convergence estimates for
recent period disappeared with disentangling of the variation in literacy and
poverty ratios with fertility rates of the states. This indicates strong connec
tion between fertility convergence and socioeconomic conditions of the
states.
Beta convergence in Socioeconomic Spectrum of major India states:
Beta convergence estimates for total fertility rates across the socioeconomic
strata12 socioeconomic groups (Scheduled caste, Scheduled tribe, Other
backward caste, Other caste, Hindus, Muslims, Other religion groups, Poor
est, Poorer, Middle, Richer and Richest economic groups) among the 15
states (15*12=180 cases) indicate evidence of divergence (b= 0.02566,
p<0.870, S= 2% per annum) for entire period under observation (19922006).
However, shorter interval based assessments of convergence indicated that
fertility rates of socioeconomic groups across states converged during initial
phase of 199299 (b= –0.82391, p<0.004, S= 29% per annum); while fertili
ty rates across socioeconomic spectrum of the states diverged (b= –0.23388,
p<0.334, S= 13% per annum) during the recent phase of 19992006. Overall,
b convergence estimates for socioeconomic groups across the states suggest
that evidence of convergence in earlier phase being replaced with divergence
in fertility rates in the later phase.
3.3.3 Convergence clubs
The fertility rates may not converge as strongly and consistently across
all the states but may converge more strongly in a subset of states of a partic
ular region. In India, it may be possible that the demographically advanced
south Indian states (club) are converging faster and earlier than the north India
Page 15
PERIANAYAGAM AROKIASAMY– SRINIVAS GOLI
79
states. Therefore, to examine such possibilities, we incorporated south Indian
regional dummy variable in the conditional b convergence model to estimate
fertility convergence among the south India states. Table 2 provides the
results of fertility convergence estimates among the south Indian states. The
negative b coefficient (b= –0.43388, p<0.400) for south India states demon
strates evidence of fertility convergence during the longterm phase of 1981
2009. However, in the south Indian states also fertility diverged during the
early phase of 198191 (b= 3.4064, p<0.227) but converged strongly during
the last two decades. Moreover, the speed of convergence was much more
pronounced during 19912001 (b= –1.9237) compared to 20012009 (b=
–0.0196).
Table 4 also presents convergence estimates for the club of south Indian
states across socioeconomic spectrum using the three rounds of NFHS data.
Compared to the major states, the conditional b convergence estimates for the
socioeconomic crosssections of south India states suggest evidence of strong
fertility convergence (b= –1.7956, p<0.000) for the entire period of 1992
2006. However, shortterm disaggregated convergence estimates also
revealed convergence in fertility rates across the socioeconomic stratum, but
the volume of convergence was greater during the earlier period of 199299
(b= –4.8771, p<0.000) compared to the recent period of 19992006 (b=
–4.5978, p<0.000). Overall, the results indicate that fertility rates did not nec
essarily converge strongly for all major states of India but converged more
strongly for the club of south India states.
Table 3 – Conditional b convergence for Total Fertility Rate (TFR)
among the major Indian states, 19812009