Demographic Determinants of Urban Growth in India, Mexico, and the US
Leiwen Jiang1 and Brian O’Neill2
National Center for Atmospheric Research and Shanghai University
National Center for Atmospheric Research
Abstract:
Understanding how urban growth occur across countries of different economic development levels and
different phases of demographic and urbanization transition is fundamental for a variety of scholarly and
policy arenas. This research aims to systematically investigate demographic determinants of urban
population growth in India, Mexico and the US for the recent decades. Adopting a multiregional
population/urbanization projection model and spatial mapping approach, we backcast urban growth in the
three countries and carry out a decomposition analysis of historical urbanization trends to quantify the
influences of natural growth, migration, and reclassification. It reveals that natural growth is the main
forces of urban growth in all three countries; Migratory growth contributes more to urban growth in the
less urbanized India. However, in contrast to the common perception that less urbanized countries may
experience more reclassification, this research shows an opposite results, which will need further
analysis using longer time period of data.
Extended Abstract:
Understanding how and why the spatial distribution of the human population changes over time is
fundamental for a variety of scholarly and policy arenas including migration, economic development,
and environmental change. Yet to date there is no consistent theory or modeling approach addressing
the dynamic, multiscale processes that govern spatial population distribution and change that is
recognized across the scientific literature. There are very few projections of future spatial distribution
over large regions and limited analysis of historic experience. As one component of a major collaborative
effort in exploring multiscale processes affecting spatial population distribution, this research focuses on
understanding the demographic determinants of national-level urban growth, which influences large
scale patterns of spatial distribution in the broad sense.
In demographic terms, aggregate urban growth at the national level (i.e., fraction urban) are determined
by combination of natural growth (births minus deaths) within urban and rural areas, migration between
the two areas or to/from other countries, and the reclassification of land (and population living on it)
from rural to urban (or more rarely, vice-versa). In one of the few studies that attempts to measure the
contributions of internal migration and fertility in urban growth, Chen et al. (1996) found that urban
natural increase is generally the dominant source of urban population growth, accounting for 60 percent
of growth in the study's median country, with the remaining 40 percent attributable to the combination
of migration and territorial reclassification. According to Chen et al., the rural out-migration rate
increased from 0.61 percent in the 1960s to 1.14 percent in the 1980s in the developing regions as a
whole. This trend differed significantly among the major developing regions. During the same period,
the rural out-migration rate declined steadily in Africa, but gradually increased in Asia, while in Latin
America and the Caribbean it increased first between the 1960s and 1970s but then declined. Such
trends are consistent with what is known about the economic situations and urbanization levels/growth
rates in these regions. Some scholars argue that the transition of urban growth is closely linked with
demographic transition (Becker 2007; Skeldon 2008); and in different phases of urban transition, the
impacts of natural growth and migration on urbanization vary significantly (Kelley and Williamson 1984;
Ledent 1982; Zelinsky 1972).
Studies of the determinants of urbanization typically combine the contributions of migration and
reclassification into a single term (Preston 1979; Todaro 1984; Chen et al. 1998; National Research
Council 2003), leaving a wide gap in knowledge about the determinants of urbanization trends and
therefore of large-scale patterns of spatial distribution. This gap limits understanding in a number of
ways. Migration and reclassification have different causes and involve populations of different age and
genders (Rogers 1982; Goldstein 1990) and their relative importance to urban growth changes across
various stages of the demographic and urbanization transitions (Zelinsky 1972; Ledent 1982; Kelley and
Williams 1984; National Research Council 2003; UN 2010). Moreover, the spatial distributions of
growing urban populations resulting from either migration or reclassification have distinctive patterns.
Urban growth driven by reclassification occurs in previously rural areas, more distant from existing
metropolitan centers and in smaller cities lower in the urban hierarchy (Goldstein 1990). In contrast,
migration-driven urban growth occurs closer to central metropolitan areas. Finally, multiregional
rural/urban projections usually combine migration with reclassification in deriving projections of future
changes in migration, which often results in unreasonable (usually overestimated) rural-urban migration
and therefore urbanization projections (Chan and Hu 2003; Lucas 1998). As fertility rates continue to fall
along demographic transition, one expects that migration will assume progressively greater importance
in differential urban and city growth, especially in countries whose fertility rates are converging across
cities. Hence, it will be increasingly important to effectively distinguish between reclassification (which
may arise from either fertility or migration changes) and migration as such, at all scales of analysis.
A few recent efforts have made some progress in isolating the contributions of rural-urban migration
and reclassification to urban growth in countries where data are available. Analysis of Indian data
reveals that while net urban migration alone contributed 19 percent to the urban growth of the 1960s
and 1970s, newly designated towns (reclassification) contributed around 20 percent to the urban
population growth of the 1960s-1980s and the proportion slightly increased over time (Economic
Commission for Asia and the Pacific 1995). Studies in China show that while more than 70% of urban
growth in the 1980s and 1990s was driven by rural-urban migration and reclassification combined (Lucas
1998), close to 50% of the urbanization was driven by reclassification (Chan 1994; OECD 2009). These
studies suggest that relative contribution to urban growth from migration vs. reclassification may differ
significantly across countries and stages of urbanization, and that when urbanization is rapid (like in the
1980s and 1990s of China) reclassification can play an especially significant role.
While the existing studies have shed light on our understanding of driving forces for urban growth, they
usually examine one or two demographic determinants and do not simultaneously and consistently
consider all three components of natural growth, migratory growth, and reclassification; these studies
focus on the experiences of either one single country or the group of developing countries and do not
systematically investigate the changes in countries across a spectrum of development and urbanization
levels; and methods used in these studies also differ from each other and make it difficult to have a
direct comparison between their research results. In our research, we use multistate demographic
method to study and compare all three demographic determinants of urban growth in the recent two
decades between India, Mexico, and the US. Our choice of these three countries is not only because
they are big countries in three continents across different economic development levels, and
urbanization and demographic transition phases, but also because they have adequate data for our
analysis. This research aims to reveal the recent changes in driving forces of urban growth of the
different parts of the world, and eventually contribute to understanding how and why the spatial
distribution of the human population changes over time.
Methods and Data
The previous studies of determinants of urban growth mostly use aggregate level accounting equations
that do not account for the effects of different age and gender compositions between rural and urban
areas. According to Rogers (1982), the assessment using an aggregate level projection model with fixed
rates ‘is of limited value for answering the question on whether it is natural increase or net migration
that is the principal source of urban population growth’; therefore, ‘a more realistic model that allows
the natural increase rate to change over time along with rate of net urban in-migration’ is needed. This
model should be able to disaggregate the rural and urban population by age. This research uses the
NCAR multiregional population/urbanization model, which is based on multi-state demographic
methods (Rogers 1995) conduct the analyses. While a few demographers have used multi-state
modeling to investigate the relative contributions to urbanization of natural growth and migration, they
have generally ignored the contribution of reclassification of areas from rural to urban. We estimate the
relative contribution of all three processes by using a novel combination of spatially explicit analysis with
the multi-state modeling.
A detailed description of the NCAR multi-regional population/urbanization model is contained in an
published article (Jiang and O'Neill 2009) and will not be repeated here. In simple terms, the basic
accounting strategy of the model is as follows.
Pxt11,s,r  (Pxt,s,r  Pxt,s,u  mx,s,u r  Pxt,s,r  mx,s,r u )  S x,s,r
Pxt11,s,u  (Pxt,s,u  Pxt,s,r  mx,s,r u  Pxt,s,u  mx,s,u r )  S x,s,u
t 1
t 1
where Px 1, s ,r and Px1,s,u are the population aged x+1 with sex s, in rural and urban area, at time t+1;
mx, s ,u  r is the urban to rural migration rate of age x and sex s, Sx, s, r is the survival rate for rural
population of age x and sex s.
For the youngest age group in the rural area, the formula is expressed as
49
P0t41, s ,r  ( ( Pi ,t f ,r  ASFRi ,r  Pi ,t f ,r  S i , f ,r  ASFRi 1,r )  2
i 15
P
t
0 4, s ,r
 0.5  m0, s ,u r  P0t4, s ,u  0.5  m0, s ,r u )  S 0, s ,r
t
where ASFR i , r is the age specific fertility rate of rural women, Pi , f , r is the rural female population of
age i at time t.
We use this model to carry out a decomposition of historical urbanization trends in our three
study areas to quantify the separate influences of natural growth, migration, and reclassification. Our
general approach will be to use the projection model to “backcast” urbanization. Data on fertility and
morality differentiated by urban and rural residence will allow us to identify the contribution of natural
growth. Depending on data availability, we take one of two approaches to distinguishing migration from
reclassification: (1) where data on both changes in rural/urban residence (due to both migration and
reclassification) and actual rural-urban migration are available, we estimate the migration contribution
using the multiregional model, and treat reclassification as the residual; (2) where such data on ruralurban migration are not available, we use a spatial mapping analysis to estimate reclassification, and
treat migration as a residual.
It is noteworthy that the multi-regional model used in our analysis explicitly represents gross
rural-to-urban and urban-rural migration rates, rather than a single net migration rate, allowing for
more detailed analysis of urbanization-migration, although at the cost of additional challenges in
obtaining adequate data to derive these rates (for a detailed description see Jiang and O’Neill 2009). For
data on population by age, sex and rural/urban residence, we use Census micro-sample data for each
decade of the 1990-2010 period for the US and Mexico (IPUMS) and detailed tabulations of population
for India (decadal, 1990-2010). Because Mexican data do not include rural/urban status of residence,
we construct such a variable by assuming that all households within a given district are either rural or
urban, and use national aggregate statistics on the proportion of urban population of each district to
define a threshold for categorizing a district as either urban or rural, such that implied total urban
population matches the national urbanization level.
For fertility and mortality data, we construct model schedules of age-specific fertility rates and
life tables for rural and urban populations. When sufficient data are not available (e.g. in older microsample data), we use the Coale-Demeny Model Life Table (Coale and Demeny 1966) and UN Life Table
for Developing Countries (UN 1982), assuming differential mortality across rural and urban areas (UN
1980; Chen et al. 1998). We also use the Brass Relation Model (Brass 1960, 1975) and data from other
sources (such as DHS surveys) to derive age-specific fertility rates.
For migration, there are two types of data availability. In the first, for India, there are only
tabulation data on changes in rural/urban residence, which result from both migration and
reclassification, so we can only estimate the rate for both of these processes combined. In the second,
for the US and Mexico, there are micro level data for both changes in rural/urban residence and
migration status (involving actual physical movement) during the census intervals. In this case, we can
estimate the rates for migration alone, and the rate for migration and reclassification combined. In both
cases we use the migration model schedule (Rogers and Castro 1981) to derive and smooth age profiles.
Using the multiregional model and these input data mentioned-above, we estimate the
determinants of urbanization by backcasting aggregate urbanization levels. For the US and Mexico,
where we have rates for migration per se, we project urbanization over the historical period with natural
growth only, natural growth plus migration, and natural growth plus migration and reclassification, with
the differences between the urbanization outcomes indicating the relative contribution of each process.
For India, where we have rates only for migration and reclassification combined, we estimate
reclassification by carrying out a spatial mapping analysis. We calculate the change in population in
newly designated towns and expanding urban boundary areas by overlaying detailed administrative
maps of rural/urban classification with maps of spatial population in each of the administrative units.
With this estimate of reclassification, we decompose urban population growth by projecting
urbanization over the historical period with natural growth only, natural growth plus reclassification,
and natural growth plus migration and reclassification.
Primary Results and Discussions
Our backcasting results based on one census data internal show that model results without considering
reclassification significantly underreport urbanization levels of the later years in India (1991-2001),
Mexico (2000-2010) and the USA (2000-2010), by 0.5, 1,9, and 1.8 percentage points respectively, which
are the equivalents of underreporting absolute urban population growth by 5.9 million, 2.7 million, and
4.9 million respectively (Table 1).
Table 1 Urban population growth and urbanization in India, Mexico, and the US
1991
% urb
natural growth +
migration +
reclassification
natural growth +
migration
urb pop (million)
Urban population
Plus natural
growth and
migration
India
2001
Projected Census
25.7
27.8
2000
74.4
27.3
217.7
Mexico
2010
Projected Census
77.8
2000
79.1
75.9
286.6
77.4
280.7
82.8
81
91.7
89
USA
2010
Projected Census
227.7
256.3
251.4
The analysis clearly reveals the contribution of each demographic components (births, death, in- and outmigration, to urban population dynamics, as well as the impacts of reclassification. Figure 1 displays that
natural growth is the main source of urban growth in all three countries (India 71.1%, Mexico 83.4%, and
USA 59.4%). The contribution from migratory growth is the highest (21.8%) in India at a very low
urbanization level, but the lowest (2.9%) in Mexico as a rather urbanized country. In the most urbanized
country the US, net in-migration accounts for 13.4% of total urban population growth. In contrast to the
common perception that reclassification in a low urbanized country like India should play a more
important role than in a more urbanized country, reclassification in India accounts for only 7.1% of its total
urban population growth, significantly lower than in the more urbanized Mexico (13.6%) and the US
(27.2%). This could be due to the fact that India had experienced a very low urbanization process until
very recently, and therefore had a very little urban expansion during the period of 1991-2001. We will
continue the analysis of 2001-2011 for India when the new 2011 Census data becomes available soon, and
investigate the possibility of changes in reclassification in the most recent decade when the country
started to take off in urbanization. The further analysis of sources of urban growth for the US and Mexico
for the early decades could also explore whether there is a change in the patterns of demographic
determinants over time for these countries.
This analysis will inform our understanding of the relative importance of the demographic determinants of
urbanization, and provide quantitative information on migration and reclassification that can be used in
other parts of the multiscale analysis at City level and spatial (grid cell) level. The city and spatial level
analysis could also help inform our national level analysis about the possible biases, e.g. due to data
limitations.
Figure 1 Sources of urban growth in India, Mexico, and the USA