Short-term Migration and Consumption Expenditure of Households in Rural India

Abstract

In 2007–2008, short-term migrants (STMs) constituted 4.35% of the rural workforce in India and a total of 9.25 million rural households included STMs. Using nationally representative data for rural India, this paper examines differences in consumption expenditure across households with and without a household member who is a STM. We use an instrumental variable approach to control for the presence of a STM in a household. We find that households with a STM have lower monthly per capita consumption expenditure and monthly per capita food expenditure compared to households without a STM. STMs are not unionised, they work in the unorganised sector, they do not have written job contracts, and state governments are yet to ensure that the legislation protecting them is properly enforced. This could be one of the reasons why we do not observe higher levels of expenditure in households with such migrants.

JEL Classification::

• O1
• R23

An earlier version of the paper was presented at the Asian Population Conference 2012 and we received useful comments from the participants. This paper is written as part of the “Strengthen and Harmonize Research and Action on Migration in the Indian Context” an initiative supported by Sir Dorabji Tata Trust and Allied Trusts. We are grateful to the managing editor and two anonymous referees for their constructive comments and suggestions. The usual disclaimer applies.

Notes

¹ Adivasi is a Hindi word for the aboriginal tribal groups of India.

² There is an active debate on whether the Mahatma Gandhi National Rural Employment Guarantee Scheme, under which each rural household is entitled to a maximum of 100 days of employment in a year, will impact short-term migration. The conjecture is that since employment has to be given if there is demand, this would reduce seasonal as well as distress migration. However, the debate is far from conclusive, given the absence of appropriate data to address this question with a certain level of robustness. We do not have to control for the impact of this scheme since in the period we consider for analysis the scheme had not been rolled out nationwide. Also, the average man-days of employment offered to beneficiary households was only 33 days in the period from 2006 to 2007 (Government of India, 2008).

³ In Karamba et al. (2011), a migrant is defined as an individual living outside the household. This definition has the following shortcomings: first, they do not know whether the individual migrated out of the household is part of the household or not; and second, they are unable to distinguish between long-term and short-term migrants in a household.

⁴ If any single source from the five sources contributes at least 50% of the income of the household during the 365 days preceding the survey, the household type corresponding to that source is assigned.

⁵ Poverty is a push factor but we cannot use the information as to whether or not a household is poor in the first-stage regression since poverty status is based on consumption expenditure.

⁶ The three geographical units at the subnational level are state, national sample survey region and district. In 2007–2008, India comprised 35 states and union territories, 87 NSS regions and 588 districts. The NSS region is comprised of a group of districts within the state.

⁷http://www.ifpindia.org/Built-Up-Areas-in-India-e-GEOPOLIS.html.

⁸ In the literature on migration and consumption outcomes, the authors have valid reasons not to model who in the household is a short-term migrant.

⁹ Angrist (2000) justifies the use of a linear probability model, since a probit or logit model can be used for analysis, only if the estimated model is “exactly correct”, which is hardly the case in empirical exercises.

¹⁰ All tests have been performed following Baum et al. (2003).

¹¹ Finally, comparing the actual differences in the MPCE, we find that the mean MPCE, in natural logarithm terms, is 6.57 and 6.35 for households with no short-term migrant and households with a short-term migrant, respectively. The predicted values from the IV model show that ln (MPCE) values are 6.79 and 5.14 for non-migrant and migrant households, respectively. We can see that predicted differences after controlling for various factors are higher than actual differences. In the case of ln (MPCE for food) as well as share of food expenditure, we observe a similar pattern.

¹² Elementary occupations consisting of ‘simple and routine tasks which mainly require the use of handheld tools and often some physical effort’ require skill at the first level.

¹³ Clerks, service workers and shop and market sales workers, skilled agricultural and fishery workers, craft and related trades workers, plant and machine operators and assemblers require skills at the second skill level.

¹⁴ Following the filing of a petition, the Supreme Court of India has directed the states and union territories to implement the Building and Other Construction Workers (Regulations of Employment and Conditions of Service) Act 1996 and The Building and Other Construction Workers Welfare Cess Act 1996.