Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

India: Selected Issues

Author(s):
International Monetary Fund. Asia and Pacific Dept
Published Date:
February 2017
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Inequality and Education in India: Tackling the Equitable Growth Challenge1

Despite India’s impressive growth performance and progress in eradicating extreme poverty, income inequality has been on the rise. This chapter investigates the contribution of inequality in education levels and the resulting high skill premium to this increase in income inequality. It documents important aspects of the Indian economy and educational sector using detailed household survey data. Based on these findings, a model is developed to simulate the direct and indirect effects of potential policies on inequality, schooling decisions and output. The main finding is that while targeted transfers work well in lowering income inequality, increasing returns to education has the most pronounced impact on measures of educational attainment.

Stylized Facts

1. Despite India’s impressive economic growth performance and progress in eradicating extreme poverty, income inequality has been on the rise. Over the last two decades, India’s GDP per capita (in PPP terms) grew on average by 11 percent per year. While this rapid growth was able to lift many people out of extreme poverty, income inequality has drastically increased. While Gini estimates for India diverge strongly, all of them show a clear increase of around 5 points since the early 1990s.2 The latest estimate in the Standardized World Income Inequality Database (SWIID) for the Gini coefficient places India with a value of 47.9 far above the simple world average of 37.3, and at second place in Asia, just after China.

2. Comparing India’s experience to Korea’s episode of “growth with equity” suggests that education could be part of the explanation. From the 1960s to early 1980s, Korea grew rapidly while maintaining an equitable income distribution, in contrast to the experience of other countries where there appeared to exist a trade-off between growth and inequality. The literature has attributed this Korean success to various factors, a prominent one being the simultaneous rapid expansion of educational levels and employment opportunities. Indeed, comparing Korean and Indian education achievements displays striking differences. Average years of schooling of the adult population (above the age of 25) were at similar levels in Korea in 1960 and India in 1990. However, over the subsequent twenty years they diverged, with Korea outperforming India on educational attainment. Additionally, while India has kept up with regard to attainment of secondary and tertiary education, it has mainly been the share of primary schooling that has fallen behind.

India: Development of Inequality and Poverty

(In percent)

Source: SWIID Version 5.1, PovcalNet.

Note: Headcount refers to the percentage of the population living in household with consumption per person below the poverty line. The poverty line is $1.90 per day at 2011 PPP. The Net Gini refers to post-tax and post-transfer income.

3. More recent data on dropout rates suggest that primary school attendance is improving in India. Recent IHDS data3 hints towards an improvement in primary school attendance and completion, as more children attend and complete primary school. Of the age cohort of 15-19 year olds in 2011/12, only 5.5 percent left school before completing 5th grade. However, 20.3 percent had left before completing 10th grade. The 2014 Indian National Sample Survey (NSS) asked persons from age 5 to 29 for the major reason for discontinuing education. Most respondents report leaving school because of economic or domestic activities or financial constraints, suggesting that outside opportunities and borrowing constraints seem to play a crucial role in determining educational choices. Furthermore, the share arguing that they are not interested in education could imply low quality of the offered education or lack of knowledge with regards to the returns of education.

Average Years of Education

Source: Barro and Lee v. 2.1 (2016)

Distribution of Education Levels

(In percent)

Source: Barro and Lee v. 2.1 (2016)

India: Main Reason for Discontinuing Education

(In percent)

Note: Major reason for discontinuance for persons of age 5-29 who had ever been enrolled but were currently not attending.

Source: National Sample Survey (NSS), 71st round, 2014.

4. Public expenditure on education in India is similar to peer countries, but is more directed towards advanced levels of schooling. Public expenditure on Indian education as a share of GDP has increased from 3.1 percent in 2006 to 3.8 percent in 2012, which is higher than that of Korea in the 1970s. Yet, it is slightly lower than that of China (4 percent) and the average of the rest of the BRICS countries (5.37 percent) in 2012. Furthermore, while the government of Korea was spending between 81 percent (1960) and 54 percent (1979) of their educational budget on primary education, India only spent on average around 30 percent between 1999 and 2012 with the remainder mostly going towards secondary and tertiary education.

5. Large differences in private expenditure on education translate into substantial differences in the quality of education achieved. IHDS data suggests that private expenditure on education varies substantially for different schooling and consumption levels. The top 20 percent in the consumption distribution spend 13 times as much as the bottom 20 percent on primary education, which declines to 6 times as much for tertiary education. Using the IHDS’ test of basic reading, writing and math skills for 8 to 11 year olds as proxies for educational quality suggests that these spending differences have far-reaching implications. Among the poorest 20 percent of households, 22.4 percent of children were not able to read at all, while this was at only 2.25 percent for the richest 20 percent. Even higher for both groups, 40.7 percent (10.1 percent) of the bottom 20 percent (top 20 percent) were not able to write.

India: Average Annual Education Expenditure by Level of Education

Note: Quintiles refer to consumption per household member.

Source: IHDS 2011/2012.

India: Abilities 8-11 year olds

(In percent)

Note: Quintiles refer to consumption per household member

Source: IHDS 2011/2012.

6. India’s educational distribution is characterized by a dual structure of modern and traditional enterprises and employment. The IHDS allows for the division of households along two main dimensions of economic activity: modern vs. traditional and entrepreneurs vs. workers.4 This classification suggests that more than 42 percent of households rely mainly on entrepreneurial income, most of them being traditional. For households relying on income from employment, two thirds are traditional workers. Educational attainment of the household’s main bread winner diverges substantially between the occupations. Modern workers achieve the highest mean years of schooling with 9.9 years, while traditional workers on average do not even complete primary school.

India: Occupational Distribution
Share in Total HouseholdsAverage Years of EducationAverage Income (normalized)Share of Poor Households
Modern Entrepreneur5.4%8.51006.7%
Modern Worker19.5%9.9996.3%
Traditional Entrepreneur37.0%5.74517.1%
Traditional Worker38.1%4.43826.5%
Source: IHDS, 2011/2012.
Source: IHDS, 2011/2012.

Policy Experiments

7. We develop a model to simulate the effect of different policy interventions on income inequality, schooling decisions and output. In particular, we are aiming to highlight and understand the effect of cash transfers, educational quality and industrial policy.

8. A heterogeneous agent DSGE model is developed to simulate the direct and indirect effects of potential policies on inequality, schooling decisions and output. Based on the empirical findings a heterogeneous agent DSGE model similar to Mestieri et al. (2016) is developed, in which agents can endogenously choose their occupations and their child’s educational level and investment, while being financially constrained. Time spent in school and private and public investment in education determine the level of human capital an individual can accumulate. The human capital production function allows for different returns to education by schooling level, and previous investments in education influence the returns to subsequent investments. Occupational choices allow agents to become unskilled or skilled workers or traditional or modern entrepreneurs. The model also features a government sector that directly funds education through public expenditure and supports households through subsidies. It finances itself through taxes on entrepreneurial profits and wages.

9. The model is calibrated to represent the Indian economy, using the most recent data from 2011/12. The baseline of the model is calibrated to the most recently available data as the chapter aims to evaluate the general equilibrium effects of different policies. Some of the parameters are chosen from the standard macro literature5, assuming that they are invariant across countries and time. The other parameters are obtained from the IHDS and various data sources, and by matching moments in order to represent the Indian economy during 2011/12.

10. Policy experiments compare the changes to the benchmark. The policy experiments range across various measures such as the impact of different kinds of cash transfers, raising educational quality and supporting the development of the manufacturing sector, some of which are being implemented or considered by the Indian government. All the experiments are comparisons of steady states and should thus be interpreted as long-run outcomes. Moreover, as government expenditure changes with household’s income, the generosity of the transfer program and the level of schooling, we adjust taxes accordingly to finance these changes.

11. Targeted cash transfers lead to lower income inequality, but conditioning on school attendance appears to be necessary to affect educational decisions significantly. The model includes a targeted cash transfer to households, which is a function of income received during that period. Thus, the higher the income of a household, the lower the government support it will receive until it passes a fixed threshold and does not receive any subsidy at all. The function for the baseline is calibrated using the volumes of government support received and reported by households in the IHDS data. Increasing these subsidies by substantial amounts has no effect on average years of schooling, but does increase the average level of human capital slightly. Additionally, inequality as measured by the Gini coefficient decreases and the share of the bottom 20 percent increases in consumption and income distributions. Conditioning the transfers on children’s school attendance can have large effects on average years of schooling and educational intergenerational mobility, which we define as the probability of a child to become skilled if its parent was unskilled. However, the transfer needs to be large enough as households face a trade-off between the loss of children’s earnings today and future higher earnings from higher schooling levels. In addition, despite higher average years of schooling, average human capital only increases slightly suggesting that longer school attendance does not necessarily imply higher productivity. Thus, higher public spending might be necessary to reap the benefits of expanded school attendance. Output fluctuates marginally as various channels seem to offset each other.

Increasing Targeted Transfers

Source: Author’s calculations.

12. Increasing returns to education raises output and educational achievements. The previous analysis of children’s abilities suggests that the quality of education might be a major concern, which has also been highlighted by other observers (e.g., Annual Status of Education Report, 2015). While it is difficult to define and subsequently translate improved educational quality, we interpret it as an increase in returns in the production function for education, i.e. for the same amount of time spent in school and money invested, a higher level of human capital will be achieved. The policy experiment focuses on an increase in returns for the first 8 years, which we classify as unskilled education. Increases by up to 10 percent would lead to significant increases in years of schooling and average human capital obtained, with a large drop in the share of the population not acquiring any schooling from 23 percent to 4 percent. While the effect on inequality as measured by the Gini appears to be non-monotonic, the bottom 20 percent of the income distribution manage to steadily grow their income share from 1.4 percent to 3.3 percent. The effect on output is large, and it more than doubles as entrepreneurs and workers achieve higher productivity through improved education and even decide to spend more time in school. As a caveat, the model does not allow us to model the potential costs of such a policy, which also need to be taken into account.

Increasing Returns to Unskilled Education

Source: Author’s calculations.

13. Shifting demand towards unskilled labor decreases the skill premium, but does not lower inequality. The Government of India’s initiative “Make in India”, launched in September 2014, has been aiming to “drive investment, foster innovation and develop skills” in specific manufacturing sectors of the economy.6 A variety of policies have and are being developed to achieve these goals, including the formulation of a framework to facilitate doing business, the opening up of sectors for investment and the development of industrial corridors. As these policies are geared toward manufacturing, they might be able to raise the demand for unskilled labor in the modern sector. In our model, shifting the weight from skilled to unskilled labor in the production function for modern sector entrepreneurs does decrease the skill premium for workers. However, this does not translate into lower inequality as measured by the Gini coefficient, on account of the top 10 percent gaining income share as they profit from unskilled labor replacing costlier skilled labor in the production function. Despite some wage gain for unskilled workers the income share of the bottom 20 percent remains constant at 1.45 percent. The magnitudes considered do not affect the average years of schooling and only slightly increase average human capital. If these policies were to additionally affect productivity or access to finance then wages and profits might increase more strongly, thereby raising the returns to education and the incentives to invest in it.

Shifting Demand towards Unskilled Labor

Source: Author’s calculations.

14. The above policy experiments illustrate the complex interactions and effects that need to be taken into account when designing policies to tackle inequality in income and education. Simple cash transfers do work well in decreasing inequality through redistribution, but might not have significant effects on average schooling levels. Conditioning transfers on school attendance can increase average years of schooling. However, if school quality is low it does not necessarily improve the productivity of students who cannot afford to invest much of their own money in education. Increasing the returns to education could have a strong effect on attendance, productivity and thus output by increasing the incentives to spend time and money on education, however it is probably also the most challenging policy initiative to define and implement as it goes beyond simple investment in infrastructure into more qualitative questions such as curricula, teaching and evaluation methods. Industrial policy aimed at a shift towards demand for unskilled labor alone could decrease the skill premium, but might not have significant effects on overall inequality as entrepreneurs would also gain.

References

    Annual Status of Education Report Centre2015Annual Status of Education Report (Rural) 2014”.

    AttanasioO.BanksJ.MeghirC. and G.Weber1999Humps and Bumps in Lifetime ConsumptionJournal of Business and Economic Statistics Vol. 17(1) pp. 2235.

    • Search Google Scholar
    • Export Citation

    DesaiS. and R.Vanneman2016India Human Development Survey-II (IHDS-II), 2011-12,” ICPSR36151-v3. Ann Arbor MI: Inter-university Consortium for Political and Social Research [distributor] http://doi.org/10.3886/ICPSR36151.v3.

    • Search Google Scholar
    • Export Citation

    Jain-ChandraS. and J.Schauer2017Inequality and Education in India – Tackling the Equitable Growth Challenge,IMF Working Paper forthcoming.

    • Search Google Scholar
    • Export Citation

    PovcalNet: the on-line tool for poverty measurement developed by the Development Research Group of the World Bankhttp://iresearch.worldbank.org/PovcalNet/.

    MestieriM.SchauerJ. and R.Townsend2017Human Capital Acquisition and Occupational Choice: Implications for Economic DevelopmentReview of Economic Dynamics forthcoming.

    • Search Google Scholar
    • Export Citation

    SoltF.2016The Standardized World Income Inequality DatabaseSocial Science Quarterly Vol. 97SWIID Version 5.1.

    StokeyN. and S.Rebelo1995Growth Effects of Flat-Rate TaxesJournal of Political Economy Vol. 103 pp. 51950.

1

Prepared by Sonali Jain-Chandra and Johanna Schauer.

2

World Bank’s PovcalNet reports for India a consumption based Gini coefficient of 33.9 for 2010 and 35.5 for 2012, while Solt’s Standardized World Income Inequality Database (SWIID) aiming for comparability across countries, estimates an income-based Gini coefficient of 47.9 for 2011. The Gini coefficient is an inequality measure ranging from 0 to 100, where 0 signifies that everyone has the same income (very equal distribution) and 100 implies that the richest person has all the income (very unequal distribution).

3

The India Human Development Survey (IHDS) is a nationally representative panel survey organized by researchers from the University of Maryland and the National Council of Applied Economic Research in New Delhi (see Desai and Vanneman, 2016). It has been used by the Luxembourg Income Study, the World Bank’s Povcalnet, and many other researchers (see http://ihds.info/papersusing-ihds-public-data).

4

See Jain-Chandra and Schauer (2017) for a detailed definition.

5

For example, the discount rate is taken from Stokey and Rebelo (1995) and the degree of relative risk aversion from Attanasio et al. (1999).

Other Resources Citing This Publication