Human Development

Income risks and schooling investment in rural Bihar

  • Blog Post Date17 September, 2018
  • Print Page
Author Image

Alok Kumar

University of Victoria

kumara@uvic.ca

Income and earnings are highly volatile in developing countries as a majority of the population relies on agriculture or informal jobs for their livelihood. Based on a primary survey in rural Bihar, this article seeks to quantify income risks faced by households, their effect on schooling investment, and whether the effect is different for male and female children.

 

Incomes and earnings are highly volatile in developing countries.  Most households in these countries rely on agriculture and informal jobs for their livelihood, which are inherently risky. At the same time, due to inadequate development of financial markets, these households have limited access to formal credit and insurance mechanisms, reducing their opportunities to diversify income risks (Dercon 2005). It is well-known that income risks may lead to highly persistent poverty (Jalan and Ravallion 2001), and prevent poor households from taking advantage of high-return but risky economic opportunities (Eswaran and Kotwal 1989, Rosenzweig and Binswanger 1993).

Empirical evidence shows that the returns from schooling in developing countries are relatively high (Peet et al. 2015). However, schooling attainment and investment are relatively low. One important issue is whether household income risks induce poor households to underinvest in schooling of their children. In recent research, I quantify income risks faced by poor households in rural Bihar and examine their effects on schooling investment (Kumar 2017, 2018a, 2018b). I examine whether household income risk reduces schooling investment; if the effect is different for male and female children; and whether the response of poor households to income risks is different from that of richer households.

Bihar is one of the poorest states in India with inadequate coverage of formal banking and financial sector. The educational attainment is low and the quality of education as measured by learning outcomes is very poor. A recent report finds that only 41.8% of children in grade 5 can read grade 2-level text, and this percentage shows a declining trend over time (Annual Survey of Education Report, 2016). Similar is the case with respect to learning outcomes for mathematics. 

Data

Data for the study comes from a primary household survey undertaken in 12 villages in Bihar in January-March 2017 (see Kumar 2017, 2018a and b for details). The survey was funded by the International Growth Centre (IGC). The sample consisted of 659 households with 1,365 children in the age group of 5-17 years. The survey collected detailed information on household and village characteristics and schooling indicators including school enrolment, annual household expenditure on schooling for each child, time-spent (in hours) by a child in school and time-spent (in hours) by a child studying (tuition, homework, etc.) outside school hours in a week, and participation of children in home production and market work.

One important challenge was to quantify the income risk faced by households, since it is not directly observed. Most studies use past income data to derive measures of income risk such as variance. But deriving income variance using retrospective data requires strong assumptions regarding how the agents form expectations, the information they have, and how they process the information (Dominitz 2001). An alternative approach, which does not require stringent assumptions, is to use subjective expectations, which involves directly eliciting probabilistic expectations from agents (Maski 2004, Delvande 2014). In this work, I used the alternative approach.

To elicit the perception of households regarding income risks faced by them, the survey contained a module designed to obtain information on the cumulative probability distribution of household income over the next year. Using survey data, I calculated household-specific expected future income and two indicators of income risk/variability:  variance of future income and its coefficient of variation. 

Main findings

Data show that the overwhelming majority of children in the 5-17 years age-group (89.7%) were enrolled in school, with the percentage of female children enrolled being marginally higher (90.9%) than of male children (88.3%). Majority of children were enrolled in government schools (87.1%), with significantly greater percentage of female children (91.7%) being enrolled in the government schools than male children (82.5%). The percentage of male children enrolled in private schools (15.8%) was more than double of female children (6.3%).

Data on time spent in school show that majority of children – both male and female – spent between 21-30 hours in school with the average being 26.67 hours. However, there is a significant gender difference in the average time spent by children on studying, tuition, or homework outside schooling hours. The average time-spent studying outside school hours in a week by children was 13.21 hours.  Male children spent more hours studying outside school hours (14.09) than female children (12.31). In addition, around 42% of female children spent 10 hours or less studying outside school hours; the corresponding figure for male children was 32%.

The gender difference in household schooling expenditure was particularly stark. The average annual schooling expenditure per child was Rs. 5,834, which was about 6% of average household income. The expenditure on uniform and teaching material (Rs. 2,248), private tuition (Rs. 1,546), and school fees (Rs. 1,365) were the three key components. The annual average schooling expenditure for male children was much higher (Rs. 7,505) than for female children (Rs. 4,163). In particular, the average spending on books, uniform, and related material for male children (Rs. 3,264) was over 2.5 times that of female children (Rs. 1,233). 

Overall, evidence suggests that there is a bias against female children in schooling investment of households, particularly with regard to time spent studying outside school hours, and schooling expenditure. Such gender differences are not reflected in other indicators of schooling such as school enrolment.

The analysis finds that household income risk has a significant negative effect on school enrolment, schooling expenditure, and time spent by children studying outside school hours (see Kumar 2017 for details). There is a significantly larger negative effect for low-income households as compared to higher income households. Separate analysis for male and female children shows that income risk has much larger negative effect on school enrolment, schooling expenditure, and time spent by children studying outside school hours for female children, in low-income households. 

Policy implications

These findings suggest that income risk negatively impacts schooling investment of households, especially for female children. This adverse effect is larger for poorer households. Hence, income risk can be an important reason for the persistence of low schooling achievement and outcomes in rural Bihar, particularly for female children. Government policies that aim to reduce income risks and variability in consumption of poor households, such as provision of health insurance, unemployment insurance, old age pension scheme, and easier availability of consumption credit, are likely to have a significant positive effect on schooling. Financial inclusion programmes that allow households to mitigate income risk can encourage schooling investment. 

Further Reading

  • ASER (2016), ‘Annual Survey of Education Report: Rural Bihar’, New Delhi.
  • Delavande, Adeline (2014), “Probabilistic Expectations in Developing Countries”, Annual Review of Economics, 6: 1-20.
  • Dercon, S (2005), Insurance against Poverty, Oxford University Press.
  • Dominitz, Jeff (2001), “Estimation of Income Expectation Models Using Expectation and Realization Data”, Journal of Econometrics, 102: 165-95.
  • Eswaran, Mukesh and Ashok Kotwal (1989), “Credit as Insurance in Agrarian Economies”, Journal of Development Economics, 31: 37-53.
  • Kumar, A (2017), ‘Income Risks and Investment in Schooling in Bihar’, Project Report, International Growth Centre, Project Code: 34309.
  • Kumar, A (2018a), ‘Subjective Income Expectations and Risks in Rural India’, WIDER Working Paper 2017/65; forthcoming, Journal of Developing Areas.
  • Kumar, A (2018b), ‘Subjective Household Income Risks and Schooling Investment in Rural India’, forthcoming, Journal of Developing Areas.
  • Jalan, Jyotsna and Martin Ravallion (2001), “Behavioral Responses to Risks in Rural China”, Journal of Development Economics, 66: 23-49.
  • Manski, Charles (2004), “Measuring Expectations”, Econometrica, 72: 1329-76.
  • Peet, Evan, Günther Fink and Wafaie Fawazi (2015), “Returns to Education in Developing Countries: Evidence from the Living Standards and Measurement Study Surveys”, Economics of Education Review, 49: 69-90.
  • Rosenzweig, Mark R and Hans P Binswanger (1993), “Wealth, Weather Risk, and Agricultural Investment”, Economic Journal, 103: 56-78.
No comments yet
Join the conversation
Captcha Captcha Reload

Comments will be held for moderation. Your contact information will not be made public.

Related content

Sign up to our newsletter