Why is the Rural to Urban Migration Rate in India so Low? An Empirical Analysis∗

Little research has examined the relationship between rural to urban migration, urbanization, and economic growth in India. While urbanization is a key driver for economic growth in many countries, the rural to urban migration rate in India is low, potentially due to a lowering of the gap between urbanrural health, infrastructure, employment opportunities, and economic conditions. Using panel data models, I investigate the relevant determinants of rural to urban migration at the state level in India from 1991 to 2011. Panel data analysis suggests that higher per capita state income and a lower difference between urban to rural literacy rates encourage rural to urban migration. In this context, I suggest that urban job creation, improved urban infrastructure, and management of urban poverty and income inequality are essential to promoting rural to urban migration in India.


INTRODUCTION
Developing countries such as India are going through a transformation from an agriculturebased rural economy to an industry-and services-led urban economy. Urbanization seems to be an inevitable part of the development process in many developed countries, where urbanization is defined as the share of the total population residing in urban areas. The basic idea of development through rural to urban in-migration (part of urbanization) is that rural resources become productively employed in urban areas, which results in overall economic growth and development for the developing country.
In this study, I aim to understand the factors contributing to rural to urban migration at the state level in India and reasons for low rates of rural to urban migration within some states in India. In India, interstate migration is very low for various reasons, such as caste systems, traditional values, diversity of language and culture, and the predominance of agriculture (Bhagat, 2014). The incentives to migrate within states in India depend on the utility differential between rural and urban areas. In this paper I consider whether differences between urban and rural health conditions, availability of infrastructure, employment opportunities, level of education, and economic conditions are really hindering rural to urban migration or not. I use data from 1991, 2001, and 2011, by sourcing data from the Census of India and National Sample Surveys (NSS) conducted by the Government of India.

BACKGROUND
After India's independence from England, the rate of urbanization began to rise steadily. The urban population in India was 28.53 percent in 2001 and increased to 31.16 percent by 2011. Urbanization has helped the country achieve higher economic growth and is, therefore, considered by many to be the main engine of economic growth in India (Tripathi, 2013;Tripathi and Mahey, 2017). Currently, 31.16 percent of the urban population is contributing to about 63 percent of India's GDP (Government of India, 2011). In developed countries, the urban population is a much higher percent of the total population. For example, Sweden's urban population was 85.8 percent of the country's overall population, 75.3 percent in Germany, 63 percent in Ireland, 81 percent in Canada, 90.5 percent in New Zealand, 89.4 percent in the United States, 80.5 percent in Netherlands, and 82.6 percent in Australia as of 2015 (United Nations, 2014). This indicates that India's urbanization rate is much lower than that of developed countries. 1 In India, the urban population may increase due to the natural growth of population currently in urban areas, reclassification of rural areas into urban areas, changes in the boundaries of urban areas, and net rural to urban migration. Recent census data suggests that the natural growth of population contributed about 43.8 percent to the overall urbanization, whereas reclassification and boundary changes accounted for 35.6 percent, and rural to urban migration 20.6 percent from 2001 to 2011. Among the cities, the average rural to urban migrant population rate (over 10 years) in Mumbai was 17.3 percent, 13.8 percent in Delhi, 8.8 percent Kolkata, 4.9 percent in Chennai, and 4.9 percent in Hyderabad, which is quite higher than other million-plus population cities. Overall, these numbers suggest that the contribution of rural to urban migration to urbanization in India is very low.
In India, policies have always been targeted at "checking" rural to urban migration. Mahatma Gandhi once said that India lives in her villages. Policies in the recent past also encouraged living in rural areas versus migrating to urban areas. For instance, the Green Revolution in India from 1967-68 to 1977-78, which significantly increased agricultural production and agricultural employment, kept people in the rural areas. Currently, ongoing rural employment guarantee programs aim to retain labor in rural areas and prevent them from migrating to urban areas, such as the one following the Mahatma Gandhi National Rural Employment Act (MGNREA), which launched in 2006. 2 Most recently, using static and dynamic panel data models, Tripathi (2018) found that agricultural activities have a negative effect on urbanization. Therefore, rural to urban migration in India has been discouraged over different time periods. As urban areas use resources more productively than rural areas, India needs to promote rural to urban migration for higher economic growth.
Census data also provide clues regarding potential reasons for rural to urban migration into cities and indicates the main reason for migration is employment opportunities. Migration to Mumbai, Delhi, Chennai, and Hyderabad also show the same pattern; employment opportunities in these cities dominate higher rural to urban migration. Marriage and consequent relocation of households are also cited as a cause for rural to urban migration.
The National Sample Survey (NSS) on "Employment & Unemployment and Migration Particulars in 2007-08" takes into account all mobility, including within and between states, and estimates the mobility of the Indian population to be about 29 percent during 2007-08. The migration of households was largely limited to within Indian states: 78 percent of the migrant households in rural areas and 72 percent of the migrant households in the urban areas showed their last usual place of residence to be within the same state. Figure 1 provides an outline of the state trends of rural to rural, rural to urban, urban to urban, and urban to rural migration. Rural to urban migration was the highest in Delhi at 56 percent and lowest in Himachal Pradesh, at 8 percent. Urban to urban migration was quite highest in Delhi at 38 percent, and particularly high in Goa, Mizoram, Nagaland, and Manipur. The figure shows that the rural to urban migration rate is lower than that of rural to rural migration in most of the states in India. On the other hand, developed or economically advanced Indian states, such as Delhi, Goa, and Manipur, have a higher level of rural to urban migration rate than less developed states, such as Punjab, Assam, and Bihar. The relationship between urbanization and income levels across India's states is shown in Figure 2. In Figure 2, a simple regression equation is fitted to the levels of urbanization in 2011 and the logarithm of per capita income of the states in 2011-12 at constant 2004-05 prices. This figure clearly indicates that higher levels of per capita income are associated with higher levels of urbanization, and the relationship is statistically significant (at the 1 percent level) with an adjusted R 2 of 0.50. From this initial analysis, it is clear that rural to urban migration -a key component of urbanization -is essential for higher economic growth.

REVIEW OF THE LITERATURE
The previous literature can be classified mainly into two categories: theoretical and empirical. Empirical research could be sub-classified into two categories: the factors that promote rural to urban migration and the factors that are detrimental to such migration. The forces that promote rural to urban migration could be divided into pull factors (toward the urban area due to its job opportunities and expected income) and push factors (out of the rural area). Although rural to urban migration is a necessary component in the development process, is it empirically justified or not? In other words, does rural to urban migration actually promote development?
In the context of how rural to urban migration promotes economic development, Saracoglu c Southern Regional Science Association 2021.
and Roe (2004) use a Ramsey-type growth model to explain that migration of the labor force from rural to urban areas has been an important component of the urbanization process. Zhang (2013) assessed China's rural to urban migration and its impact on economic development. The author of the study found that migration toward cities made a great contribution to economic growth and was a dominant source of an increase in urbanization. In another micro-level study, Imoru and King (2017) found that rural to urban migration contributed to the economic development of migrants in the Tamale district of Ghana.

Theoretical Studies
Various economic theories establish that structural change is an inevitable component of economic growth. In the Lewis (1954) model, shifting labor from the primitive agriculture sector to manufacturing triggers growth by generating an investable surplus. In the process of structural change, both internal and external agglomeration economies are generated. The Todaro (1969) model explained rural to urban migration as a function of the income differential that is adjusted for the probability of finding a job. The Harris and Todaro (1970) model attempted to explain the phenomenon of accelerating rural to urban labor migration despite the existence of positive marginal products in agriculture and significant levels of urban unemployment.

Trends and Patterns of Migration in India
Discussing the general trends and patterns of migration, Mitra and Murayama (2009) found that the intrastate migration rate -the rate of migration within states in India -is much higher in magnitude than the interstate migration rate -the rate of migration between states -in India. Male and female migration rates are closely interconnected, irrespective of whether they migrate from rural areas within the states or to other states (Mitra and Murayama, 2009). The social and cultural diversity in India stands as a major hindrance to population mobility (Mitra and Murayama, 2009). Banerjee (1986) suggested that the underlying objective of migration is the maximization of a family rather than individual benefits. Bird and Deshingkar (2009) explored circular migration in India. Circular migration is when poor people from low productivity areas migrate for part of the year, often on a seasonal basis, to areas with more job opportunities and wages. They found that circular migration rates are high in remote rural areas, particularly among the chronically poor. Chauvin et al. (2017) argued that India's rural to urban migration is limited by strong placebased preferences, such as those related to cast-based social networks in India. Moscona (2017) found that the green revolution has had adverse effects associated with pulling labor and resources away from other, potentially more productive, sectors.

Pull Factors Promoting Rural to Urban Migration
In the case of the United States, Arzaghi and Rupasingha (2013) argued that migration from rural to urban areas improves an individual's economic choices and opportunities, because migrants can look for opportunities in both the origin and destination locations. Iversen (2006) presents a dynamic theory of rural to urban migration that includes social networks, including the social cohesion of origin communities and their links to urban workplaces. According to McCatty (2004), who examined the need for rural to urban migration in developing countries, migration takes place in these countries due to job opportunities, education, medical facilities, high per capita income, transport facilities, and higher standards of living.
Haurin ' (1980) model evaluated the impact of a favorable climate (which is a site-specific factor) on the city population (along with the price of housing and price of factors) and found that a relative increase in the climate differential would induce migration toward the improved area. Urban areas, which are better endowed with location-specific attributes, attract more population and grow more when compared with areas with less favorable climate. A theoretical analysis by Issah et al. (2005) confirmed that the presence of infrastructure can be a significant factor in migration decisions.
In the context of India, based on the primary survey and using the probit model, Sridhar et al. (2013) found that an increasing level of education of the migrants acts as the main pull factor for migration in Bengaluru. Shanthi (2006) examined the extent of employmentoriented migration of females in India and the interstate variations in its magnitude using National Sample Survey data in 1999-00 at the household-level. The author found that the "independent migration" of females is increasing in response to employment opportunities in export industries, electronic assembling, and garment units. Ghate (2009) highlighted the case of Oriya migrants to the cities in Gujarat. He argued that the job opportunities in Surat's power loom industry pooled many Oriya migrants from a few contiguous blocks in the Ganjam district in Orissa. Banerjee (1984) used probabilistic migration models which assume that search for urban jobs is entirely an urban-based activity and that employment in free-entry activities is a transitional phase, during which migrants are actively searching for formal sector employment. Based on the sample survey, Banerjee found that the Delhi experience does not fit the migration process postulated in the probabilistic models. One-half of the surveyed migrants had moved to Delhi after lining up specific jobs. Mitra (2019) found that the migration rate in urban areas and the urbanization level are positively associated but moderate in magnitude. With an increase in the city size, the migration rate rises primarily due to better employment prospects in large cities from agglomeration effects.

Push Factors Promoting Rural to Urban Migration
While investigating the contribution of push factors in rural to urban migration, McCatty (2004), in the context of developing countries, found that political disputes, strikes, fighting with neighbors, lack of transportation, illiteracy, natural disaster, and bullying push rural people to urban areas.
In the context of India, Sridhar et al. (2013) found that the lower level of education of the migrants acts as a push factor for migration in Bengaluru. Agasty and Patra (2013) examined the determinants of rural to urban migration in the Indian state of Orissa and found that low income, high poverty, high unemployment, lack of rural job opportunities, and lack of farmland are the reasons for rural to urban migration. Deshingkar (2004) suggested that c Southern Regional Science Association 2021. individuals may be pushed to migrate because of debt, poor access to credit, declining access to common property resources, or commodity price crashes in India. Banerjee and Kanbur (1981) used a probabilistic migration model to examine internal migration in India and found that migration tends to rise at first but then falls as rural income rises.

Factors Detrimental to Rural to Urban Migration
Considering the factors that are detrimental to rural to urban migration in India, Akram (2015) found that an increase in per capita Net State Domestic Product tends to decrease the number of out-migrants from the rural areas of that state. However, an increase in the proportion of the population living below the poverty line, a higher proportion of Scheduled Castes in the population, and a higher illiteracy rate in the rural area of the state tend to decrease rural to urban labor migration from that state to other states. Bhagat (2014) argued that migrants with low education and skills, given the seasonal and temporary nature of their employment, are more vulnerable and subject to various kinds of exclusions in urban areas. Migration is treated more as an issue of governance rather than one of development in developing countries like India. Deshingkar (2004) argued that the Watershed Development (WSD) program, which has been introduced to reduce rural to urban migration in India, needs to be reexamined. Specifically, research has synthesized the available evidence to show that the relationship between migration and WSD is complex and depends on a variety of factors ranging from rural to urban wage differences, personal aspirations, and education levels. Therefore, more empirical research is urgently needed in this area. Most importantly, Jacob (2008) assessed the impact of NREGA on rural-urban migration in the Villupuram district of Tamil Nadu. He found that given the decent employment created by NREGA in their villages most people, both men and women, would not prefer to migrate. However, Ahuja et al. (2011) found that despite being a source of employment, MGNREGA has not been able to contain migration from the developed region because of higher market wage rates in destination states like Haryana. Figure 3 explains all the push and pulls factors behind rural to urban migration which c Southern Regional Science Association 2021.
has been discussed in the review of the literature. The literature review suggests that though there are several descriptive and micro-level studies, systematic statistical analyses to investigate factors that increase rural to urban migration at state levels are limited. Different time dimensions have not been taken into account on migration studies in India. This has led to an overall lower rate of urbanization in India. This study tries to fill this research gap. The research outcomes of this analysis will help policymakers to come up with appropriate measures to promote urbanization through higher rural to urban migration at the state level in India.

EMPIRICAL FRAMEWORK AND RESULTS OF THE ESTIMATION OF DETERMINANTS OF RURAL TO URBAN MIGRATION IN INDIA
The econometric model of state-level rural to urban migration takes the following form: where y it is the total number of migrants or percentage of migration from rural to urban areas and X it is a set of explanatory variables. η i is the unobserved time-invariant, state-specific effects; δt captures a common deterministic trend; it is a random disturbance assumed to be normal, and identically distributed with E( it ) = 0; V ar( it ) = σ 2 > 0. The state-specific percentage of urban migration is defined as the total number of migrants from the rural areas to urban areas of a particular state with the duration of residency, of less than one year to more than 10 years, divided by the total urban population of that state. The Census of India provides rural to urban migration data from less than one year of residence to more than 10 years of residence. We consider all the categories years of residence. More specifically Census of India provides data for the following categories: (1) Duration of residence less than 1 year, (2) Duration of residence 1-4 years, (3) Duration of residence 5-9 years, and (4) Duration of residence 10 years and above. In this paper we consider all of these types of migration when defining a migrant. That is, a rural migrant to an urban area may fall into any of the four categories listed above.
On the other hand, the total number of migrants of a state is defined by the total number of migrants from the rural areas to urban areas of a particular state with the duration of residency of less than one year to more than 10 years. Independent variables are mainly considered based on the differences between rural and urban work opportunities, medical facilities, availability of infrastructure, economic conditions, and educational facilities. The consideration of independent variables is mainly based on the review of the literature and the availability of data. The description and source for the independent variables are listed in Appendix A1. The panel data estimations consider data from three Census periods, 1991Census periods, , 2001Census periods, , and 2011. We consider 28 states and 7 union territories of India for the analysis. 3 Table 1 presents the descriptive statistics of the variables used in the panel data model. The standard deviations for the average monthly per capita consumption expenditure and the labor force participation rate are very high compared to the other variables that are considered indicating that the variability among these two variables is high. Table 2 presents the result of the determinants of state-level total rural to urban migration based on equation (1) using panel data methods. To improve the model fit, I ideally would use the logarithmic transformation of the dependent and independent variables. However, these variables measure the differences between urban and rural migrations and some of them have negative values, which means I cannot use a logarithmic transformation. 4 Regression models 1 and 6 show the estimates of the full model, which includes all variables for the maximum number of available observations. Regression models 2-5 report the estimation of the parsimonious model. First, I conduct an F-test to determine whether the fixed effect model or the pooled model is better. The significance of F-test compels us to use the fixed effect model over pooled model. Next, I perform the Breusch-Pagan Lagrange multiplier (LM) test using STATA command xttest0 to check whether to use a random effect panel data model or a simple OLS regression. The significant value of the chi 2 of the LM test validates the use of the random effect panel model. However, the significant value of chi 2 of the Hausman test validates the choice of the fixed effect model over the random effect model for the estimation. As the time points of data are very small, I did not find any problem of cross-sectional dependence/contemporaneous correlation and serial correlation in the estimation. However, I found there was heteroskedasticity present when I tested for it using the STATA command xttest3. To address this problem, I used Huber/White or sandwich estimators by using a 'robust' option to obtain heteroskedasticity-robust standard errors. The robust fixed-effect models perform well, explaining up to 74 percent of the sample variation in India's state-level total rural to urban migration in regression 1. The statistically significant values of the F-statistics indicate the overall significant level of the Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. estimated regression models. Regression models 1-5 use the log of the total number of rural to urban migrants at the state level as a dependent variable. Regression model 5 shows that lower differences between urban to rural infant mortality rates reduce total rural to urban migration. The result is statistically significant at a 5 percent level. A 1 percent reduction in the infant mortality rate reduces total rural to urban migration by 3 percent on an average. The infrastructure variables, i.e., the percentage of households having electricity and safe drinking water, also indicate that the lower the difference between urban and rural areas, the lower the rural to urban migration rate. For instance, a reduction of 1 percent urban to rural electricity connectedness reduces the total rural to urban migration rate by 2.5 percent in regression model 2. Employment opportunities, which are measured by the labor force participation rate, show that a 1 percent reduction in urban to rural area decreases total rural to urban migration by 0.4 percent in regression model 3. Average monthly per capita consumption expenditure, which represents the economic conditions, indicates that a reduction in the difference between urban to rural areas reduces total rural to urban migration in the state level of India.
Regression model 1 shows that a reduction in the urban to rural poverty headcount c Southern Regional Science Association 2021.
ratio reduces the total rural to urban migration rate. The result is statistically significant at the 5 percent level. In particular, a 1 percent reduction in the urban to rural poverty ratio shrinks total rural to urban migration by 1.5 percent. A reduction in the extent of inequality between urban and rural areas also reduces total rural to urban migration in India. In contrast, the level of education, which is measured by the literacy rate, shows that the reduction in the difference between urban and rural areas increases the total rural to urban migration. This result indicates that rural to urban migration mainly occurs due to a higher level of educational attainment. The results support the findings of Sridhar et al. (2013) and Agasty and Patra (2013). A higher level of per capita net domestic product also encourages rural to urban migration. In particular, a 10 percent increase in per capita net state domestic product increases rural to urban migration by 8.1 percent in regression model 2. The result supports the findings of Arzaghi and Rupasingha (2013). Finally, a higher number of agriculture workers indicate that higher opportunities in agriculture reduce rural to urban migration. This result supports the impact of the green revolution and MGNREA in rising agriculture employment and production, reducing rural to urban migration (Moscona, 2017). Finally, regression model 6 uses the log of the percentage of the rural to urban migration rate as a dependent variable. The coefficient estimates from this model suggest that a lower difference between the poverty rates of urban and rural areas reduce the rural to urban migration rate. However, a higher per capita net state domestic product encourages rural to urban migration. The results are consistent with earlier coefficients discussed in regression models 1-5.

DISCUSSION
The state-level analysis suggests that the improvement in health, infrastructure, employment, education, and economic conditions deter rural to urban migration. However, it is important to mention that a higher level of urbanization is required for a higher level of economic development. Therefore, it is necessary to relocate rural resources to urban areas for higher productivity and higher economic growth. The following discussion mainly traces how urban areas or cities can attract more resources from rural areas.
The majority of the households migrated for employment purposes. It is estimated that about 55 percent of the rural households and 67 percent of the urban households migrated for employment-related reasons only in 2007-08, based on information from the 64 th round of the National Sample Survey. City-level employment has a powerful effect on rural to urban migration. The 2011-12 National Sample Survey calculated that only 56 percent of urban males and 16 percent of urban females were in the labor force. The unemployment rate for urban males was about 3 percent, and about 5 percent for urban females, which was higher than the rural unemployment rate of 2 percent for both males and females measured by the usual status. As per the information available on the official website of the Centre for Monitoring Indian Economy (CMIE), nearly 31 million unemployed Indians are looking for jobs. In February 2018, the urban unemployment rate increased to about 6.84 percent (Times News Network, 2018b). This picture indicates that India has failed to create enough urban jobs to attract people from rural to urban areas.
Indian cities and towns lack quality infrastructure and services. The Report on Indian Urban Infrastructure and Services (Government of India, 2011), which deals extensively with this phenomenon, has emphasized the importance of the delivery of urban public services such as drinking water, sewerage, solid waste management, roads, and street lights for the sustainable urban development in future in India (Tripathi, 2018). Only 64 percent of the urban population in India is covered by individual connections and stand posts. 5 The duration of water supply in Indian cities ranges only from 1 to 6 hours per day. Government of India (2011) indicated that, out of the 5,161 cities and towns in India, 4,861 do not have even a partial sewage network and 18 percent of urban households do not have access to any form of toilets and people defecate in the open. Public transport accounts for only 22 percent of urban transport in India. Therefore, a lower level of urban infrastructure could not bring more people from rural areas.
Inequality measurements based on NSS consumption expenditure data shows that urban inequality in India is higher than rural inequality. Urban inequality increased by about 15 percent in the period from 1973-74 to 2011-12. On the other hand, as per the poverty headcount ratio measured by the Tendulkar methodology, it has declined from 25.7 percent (i.e., 807.6 lakh) in 2004-05 to 13.7 percent (i.e., 531.2 lakh) in 2011-12. However, there is a disagreement on how poverty and inequality are measured in India. Furthermore, India's large cities entail expensive living costs. For example, Mumbai ranks 55 th among 209 cities when measuring the comparative costs of more than 200 items such as housing, transportation, and food and clothing (Times News Network, 2018a). Therefore, substantial reductions in poverty, inequality, and living cost have not been achieved by the Indian government to pull rural people to urban areas.

CONCLUSION AND POLICY IMPLICATIONS
This paper investigates the relevant determinants of rural to urban migration at the state level in India based in 1991India based in , 2001India based in , and 2011 Census and National Sample Survey data. Fixed effect panel data models are used for the analysis.
The descriptive analysis shows that India's urbanization rate is much lower than that of other developed countries. A natural increase in the urban population is one of the main sources of the increase in the urban population in India. Net rural to urban migration only accounted for about 21 percent of the change between 2001 and 2011. The city-level data shows that the main reasons for rural to urban migration is employment opportunities and marriage.
The state-level panel data analysis shows that economic conditions measured by per capita net state domestic product and the lower difference between urban and rural literacy rates increases rural to urban migration. However, a lower difference between urban and rural areas, i.e., improvement in health (measured by infant mortality rate), infrastructure (measured by electricity connection and drinking water), employment opportunities (mea-sured by labor force participation rate and number of rural agriculture worker), and economic conditions (measured by poverty, inequality, per capita consumption expenditure) reduces rural to urban migration in India.
Considering the main question for this research, "Why is the rural to urban migration rate in India so low?" based on the analysis, I suggest that India has failed to create enough urban jobs, provide substantial infrastructure facilities to the urban dwellers, control the cost of living in the urban areas, and reduce urban poverty and inequality in order to promote rural to urban migration. I am of the view that rural to urban migration is a necessary part of the development process and does not necessarily have to result in an adverse impact on rural areas. Several rural policies such as NREGA have tried to check rural to urban migration, but even pull factors did not attract people from rural to urban areas.
It is quite obvious that a higher level of rural to urban migration is needed for economic development in India. In rural areas, the majority of the population depends upon agriculture, and the higher dependence on farming leads to disguised unemployment in rural areas. If the disguised unemployed population is relocated to urban areas, then the supply of labor and demand for consumer goods in urban areas would increase. This would, in turn, lead to more production, a higher level of economic activity, and higher per capita income. The number of job opportunities in the cities will also increase in this process. This increase promotes investments, which would, in turn, lead to further economic growth. So, the economic growth in India can be catalyzed through increased urbanization resulting from rural to urban migration. Therefore, I suggest the following policies. First, India should increase job opportunities in urban areas to enable a higher rural to urban migration. Second, the level of urban poverty and urban inequality has to be addressed in order to encourage migration from rural to urban areas. Third, basic urban infrastructure facilities such as road, electricity, and educational institutions need to be increased to facilitate investment-friendly urbanization to promote rural to urban migration. Finally, migrants should be able to get higher paid jobs so that they can survive in big cities with higher living costs. However, how cities or towns can attract more residents from rural areas and consideration of other variables such as family income diversification, social networks, and caste affinities, and employment growth to explain rural to urban migration forms the topic of future research. Also, establishing a causal link between economic conditions and employment opportunities and rural to urban migration is left for further scrutiny.