Pattern and Determinants of Household Expenditure on Higher Education: Evidence from Rural Odisha

Abstract

This chapter examines the variability of household expenditure on higher education in rural Odisha (one of the backward states of India) and its relationship with individual, household and institutional factors. This chapter uses the data collected through a student survey in two tribal dominated districts of Odisha (Mayurbhanj and Keonjhar) in 2016–2017. In total 563 scheduled caste, scheduled tribe and other backward class students pursuing their undergraduate and postgraduate courses in 19 colleges are surveyed for the study. The findings suggest that the annual average household expenditure on higher education among marginalized sections of the society in rural Odisha is approximately 30% of the annual family income. Also, students enrolled in government and aided colleges have spent more than the students enrolled in unaided higher education institutions and interestingly; this difference is largely due to the difference in the payment of non-fee items such as private tuition, food and accommodation, transport, Internet, etc. On an average, students pay only about 5% of their total expenses in higher education as fees per year and rest on non-fee items. The pattern of household spending on higher education varies significantly between hostellers and day scholars. As expected, students belonging to poor households have invested less on higher education than the households with better income. Results also indicate pro-male bias in household spending on higher education in rural Odisha.

The chapter is based on an ICSSR-sponsored study titled ‘Effectiveness of Select Scholarship Schemes for the Improvement in Access and Retention of Scheduled Caste and Scheduled Tribe Students in Odisha’.

Introduction

The Indian higher education (HE) sector has undergone massive expansion, though not uniformly, during the seven decades following independence. However, looking at the growth pattern of HE reveals that in post-1990s, there is greater intervention of the private sector in providing higher education in India and simultaneously the role of the state towards this sector has become quite insignificant. The growth in the number of private universities established during the last 5 years is unprecedented as out of the total 235 state private universities, 168 have been established after the year 2010 (UGC 2016). There are more than 78% colleges running in private sector (aided and unaided taken together) that caters more than two-thirds of the total enrolment. It is argued that the growth of higher education, particularly in the post-liberalization period, is a market-mediated process facilitated mostly through private institutions (Varghese 2015).

In the early 1960s, public funding and philanthropic contributions for higher education were the major part of the resource to this sector in India, and the contribution from private sources in terms of tuition fee and other payments from students were negligible. The trend has shifted towards private funding of higher education, particularly with the implementation of the New Economic Policy of 1991. The higher education in India is moving towards the conception of private good largely financed by households (Panchamukhi 1990; Mathew 1996; Rani 2004; Chakrabarti and Joglekar 2006; Indira 2006; Chattopadhyay 2007; Prakash 2007; Varghese 2013). In the process, apart from increasing student fees in private HEIs in India, many public universities have initiated efforts to generate additional resources through the offering of ‘self-financing courses’ where the major part of the cost is borne by students. Several studies in India reveal that in the post-1990s, households bear a significant proportion of the costs in the form of fee (particularly tuition fee) and non-fee expenditures incurred on purchasing books, stationery, uniform, conveyance, private coaching and other necessary expenses related to their children’s higher education compared to the pre-1990s. However, there is less evidence on the changing pattern of household spending on higher education in recent years, though few studies are available in the domain of school education (e.g. Tilak 2002; Kingdon 2005; Azam and Kingdon 2013). Also, there is hardly any study on this in the context of Odisha, a state with a low level of socioeconomic and human development. More importantly, there are very few studies on household costs of higher education in rural areas and also with a focus on marginalized sections of the society such as scheduled castes, scheduled tribes and other backward classes. Considering these research gaps, this chapter examines the variability of household expenditure on higher education among the students of lower social groups in rural Odisha and its relationship with their individual, household and institutional charateristics. This chapter uses the data collected through a student survey in two tribal dominated districts of Odisha (Mayurbhanj and Keonjhar) in 2016–2017.

Past Studies

A considerable body of research suggests that the amount of household investment on education differs significantly with their socioeconomic settings, particularly by gender, family income, parents’ education and location of the household (rural-urban). Past studies that have discussed these four important factors in detail are reviewed briefly in this section. A theoretical analysis of household decision-making in the investment in education by gender is discussed in Pasqua (2005). Circumstantial evidence for the existence of pro-male bias in household spending on education has been documented in many countries across the world, including India (Panchamukhi 1990; Subramaniam and Deaton 1991; Li and Tsang 2003; Gong et al. 2005; Kingdon 2005; Lancaster et al. 2008; Zimmermann 2012; Masterson 2012; Saha 2013; Iddrisu et al. 2018). Preference of the households to invest in the education of boys rather than girls is widely prevalent in India and such difference widens further in case of higher education (Chaudhuri and Roy 2006; Kambhampati 2008; Kaul 2018). Many studies have confirmed that the variation in household investment in education by gender is due to the parents’ preference for better quality education for boys (by investing more) over girls (Aslam and Kingdon 2008; Himaz 2009; Azam and Kingdon 2013; Saha 2013).

A positive relationship between household income and investment in education has been found in many studies (King 1998; Acevedo and Salinas 2000; Psacharopoulos and Mattson 2000; Urwick 2002; Psacharopoulos and Papakonstantinou 2005; Tansel and Bircan 2006; Omori 2010; Shafiq 2011; Acar et al. 2016). Examining the determinants of household expenditure on school education in rural India, Tilak (2002) has found that households tend to spend more on education with the increase of their income, the value of the income elasticity coefficient being 0.20. Educated parents (also other educated adult members of the household) are more aware of the benefits of education and hence spend more on it, which has been established in many studies, both in India and elsewhere (Kanellopoulos and Psacharopoulos 1997; Psacharopoulos and Mattson 2000; Tilak 2002; Dang 2007; Omori 2010; Masterson 2012; Saha 2013). Saha (2013, p. 233) has reported that the higher the educational level of the parents/guardians, the greater is the spending on education of their offspring in India. In a recent study, Chandrasekhar et al. (2019) have found that rural households spend 15.3% of their total expenditure on average in higher education while it is 18.4% inn urban areas. However, there are limited studies that examine the pattern and determinants of family spending on rural HEIs and socially disadvantaged students like scheduled castes, scheduled tribes and other backward classes and also in the context of Odisha. This study examines the variability in household expenditure on higher education in rural Odisha by including a set of factors (individual and household characteristics, students’ academic background and institutional factors) that are expected to influence household’s decision to invest in it.

Data and Methodology

The study uses the data collected from the primary survey conducted in 19 higher education institutions (HEIs) located in two tribal dominated districts (Mayurbhanj and Keonjhar) of Odisha in 2016–2017. The survey covers 563 students (169 SC, 353 ST and 41 OBC) pursuing their undergraduate and post graduate courses in these two sample districts¹. The representation sample students by type of institution, gender, caste, religion and region are provided in Table 10.6 in Appendix. Majority of the students (95%) surveyed are from undergraduate courses (three-fourths being from B.A. programme only), as in most of the HEIs there have no postgraduate courses except North Orissa University and MPC Autonomous College. Out of the total respondents, 48.5% are from aided colleges, 29.5% from unaided HEIs and 22% from government colleges². Around three-fourths of students surveyed are female.

The pattern of household expenditure on higher education in rural Odisha is discussed by students’ individual characteristics, household and institutional factors. Two separate household expenditure functions are estimated using Ordinary Least Square (OLS) technique to analyse potential factors determining household expenditure on higher education: first, for total household spending on higher education (fees and non-fee items) and second, the household spending on education that includes only non-fee items. Examining the factors determining household spending on non-fee items is attempted separately as it constitutes a major share of the total expenditure on higher education. Also unlike fees, the expenses on non-fee items are elastic to income/needs of the students after enroling the HEIs.

Pattern of Household Expenditure on Higher Education

The survey results show that the total expenditure per student per annum is about Rs. 61,490 which constitutes approximately 30% of the annual family income. Of the total household expenditure on HE, students spend only about 5% on fees and the rest on non-fee items like food and accommodation, textbooks, transport, tuitions, computer classes, mobile and Internet (Table 10.1). Amongst the non-fee expenditure, major proportion of expenses are made in food and accommodation (29.43%) followed by textbooks (19.31%), private tuition or coaching (18.46%), computer class (9.41%), transport (8.99%), mobile (4.75%) and Internet (4.57%). Households have spent a reasonable share of their income on private tuition or coaching. It is observed from the interaction with the students that majority of them go for private tuition as the quality of teaching in the colleges is very poor. As the students belong to socially backward communities and also in many cases being first generation learners, they need more support to perform better in class and also to continue with higher education successfully. Being from socially disadvantaged strata of the society, students have difficulty in learning and completing their courses. Interestingly, fee paid for private tuition or coaching is relatively high for science streams, followed by commerce and then arts.

Table 10.1 Annual per student household expenditure on higher education

Often, it is argued that the household expenditure on non-fee items (particularly on food and accommodation) varies substantially between hostellers and day scholars, while fees paid by them are more or less similar. This is also found to be true in this study as students residing in the hostel spend Rs. 19,498 more on non-fee items such as food and accommodation, textbooks and study materials, transport, private tuition/coaching, computer class, mobile and internet annually than the day scholars³. Per student annual household expenditure on non-fee items was Rs. 60,974 for hostel students and Rs. 41,476 for day scholars. Interestingly, students residing in the hostel spend more than half of the total non-fee expenses on food and accommodation. Also, the expense of hostellers on this is more than two times of the expenses by day scholars⁴. As expected, day scholar spends about Rs. 250 per month on transport which is close to three times the expenses incurred by the students residing in hostels. This reveals the expected trend as day scholars commute to the college from home and henceforth spend more on transportation. Similarly, expenses on mobile and Internet are more for the students residing in the hostel as compared to day scholars as they wish to remain in contact with their families regularly (Table 10.2).

Table 10.2 Annual per student non-fee household expenditure on higher education for hostel students and day scholars

The pattern of household spending on higher education by type of institutions reveals few important points. First, students pursuing their course from government and aided HEIs have spent more than the students enrolled in unaided higher education institutions, and interestingly, this difference is largely due to the difference in the payment of non-fee items such as private tuition, food and accommodation, transport, Internet, etc. (Table 10.3). This finding does not go in line with the results of many other studies which show that students attending unaided/private HEIs spend more than the government and aided institutions. It is largely because the binary between government and private HEIs discussed often in the literature is not the same here. The unaided and self-financing colleges in rural Odisha (included as sample in this study) largely offer undergraduate courses on liberal arts, basic sciences and commerce. The total cost of attending these colleges is comparatively less than the government and aided colleges as these are located in remote rural areas and very close to the vicinity of the students, and therefore households do not spend much on non-fee items such as food, accommodation and transport. The private colleges also not charge much on fees due to the fee regulation policy of the state government and also this will be a major barrier for getting students in the college. These colleges are providing higher education for long (in some cases for the last 25 years) with the expectation that they will be converted to aided colleges that will help them to get financial support from government, particularly to cover teacher’s salary and fund for infrastructure. The conversion of colleges from unaided to aided is a practice followed in past by the government of Odisha, though it is not smooth and often teachers go for long strikes to put their demand.

Table 10.3 Annual per capita household expenditure on higher education by type of institution, gender, caste and annual family income (in Rs.)

The analysis finds that male students spend more on higher education than female students on fees. Also, in non-fee items there exists pro-male bias for the day scholars. However, female students residing in hostels spend more (Rs. 62,048) than male students (Rs. 55,113) on non-fee items. The annual per capita household expenditure on higher education is substantially higher among OBC students followed by SC students and ST students, and there is not much variation in the pattern of spending between hostellers and day scholars (Table 10.3). As expected, the annual average household expenditure on higher education increases with the increase in the annual income of the family and it is true for both fee and non-fee items except some variations between lower income and lower middle-income groups⁵.

Determinants of Household Expenditure on Higher Education: OLS Results

This section estimates two separate household investment functions on higher education using the OLS technique. The first model considers the total household expenditure (fees and non-fee items), while the second model includes the spending on non-fee items. The OLS equations used for the estimation are as follows:

$$ lnEducost 1={\alpha}_1+{\beta}_i{X}_i+{\varepsilon}_1 $$

(10.1)

$$\mathit{lnEducost}\mathit{2}={\alpha }_{\mathit{2}}+{\gamma }_i{X}_i+{\epsilon }_{\mathit{2}}$$

(10.2)

Where,

• lnEducost1 = Natural logarithm of annual total household expenditure on higher education (expenses on fees and non-fee items taken together)

• lnEducost2 = Natural logarithm of annual household expenditure on non-fee items of higher education

• α₁ and α₂ = intercept terms

• β_i and ɣ_i = regression coefficients that measure the influence of explanatory variables on the household expenditure on higher education

• X_i = explanatory variables

• ε₁ and ε₂ = error terms

The explanatory variables used in the regression are broadly categorized as follows: individual characteristics, household factors and institutional factors. Several other determinants of household expenditure on education like household size, household budget for items other than education, opportunity cost of studying higher education, etc. are not considered in the analysis because of data limitations. The notation, definition and summary statistics of the explanatory variables used in estimating household expenditure functions on higher education are presented in Table 10.7 and Table 10.8, respectively, in Appendix.

The issue of gender bias in household expenditure on education has been a topic of much research, particularly in the context of developing countries. The regression results show that female students spend 14.8% less than the male students on higher education in rural Odisha. The pro-male bias in household spending is more visible for non-fee items (lnEducost2) as the value of the coefficient is 15.3% (Table 10.4). Though in some households parents send both sons and daughters to colleges, they prefer better quality education for boys (by investing more) over girls, and it is more so in rural settings and also among marginalized sections of the society. Several other studies, both in India and outside (Panchamukhi 1990; Kingdon 2005; Lancaster et al. 2008; Jensen 2012; Zimmermann 2012; Saha 2013), support this finding.

Table 10.4 OLS estimate of the determinants of household expenditure on higher education in Odisha

Though some studies have found that students from lower social class (SCs and STs) have significantly less chance of attending higher education and it is particularly low in professional courses (Azam and Blom 2009; Chakrabarti 2009), there is hardly any study that shows the variation in household investment on higher education by caste, particularly in the context of Odisha and in rural areas. The OLS result shows that SC and ST students spent around 21% and 38% less on higher education respectively as compared to OBC students in Odisha. More or less similar picture is also visible in the household spending on non-fee items. The difference in the spending may be larger if compared with the general category students whose data is not collected in the survey as the study was on marginalized sections of the society. Gangopadhyay and Sarkar (2014), using household-level data from West Bengal, find that SC households spent significantly less on private coaching of their children compared to other households, and this finding is consistent with the results of this study.

The regression coefficient for the annual income of the family (lnFamily_income) is found to be positive and statistically significant in both the models, with a marginally higher effect on non-fee items than total household expenditure on higher education. Results support the findings of many other studies which reveal that rich households spend more on the education of their children than poor households (King 1998; Acevedo and Salinas 2000; Tilak 2002; Acar et al. 2016). Quite surprisingly, parents’ education (father and mother education taken separately in the equation) did not come out to be statistically significant in explaining the variation of household spending on higher education in Odisha, though the coefficients give expected sign. This may be due to the lack of heterogeneity in the parents’ educational pattern, as close to two-thirds have attended below the primary level of education, and a similar pattern is also found among mothers.

Students enrolled in government and aided higher education institutions have spent more (on both lnEducost1 and lnEducost2) than the students of unaided institutions. More clearly, students from government institutions spent 26% more and unaided institutions spent 15% more on higher education than the students studying in unaided institutions. Both the coefficients are statically significant at 1% level of significance. Similarly, students of government and aided institutions spent around 32% and 18% more respectively on non-fee items as compared to unaided institutions. The finding here does not go with the common understanding that students of unaided/private HEIs usually spend more as compared to the government institutions. Therefore, it is important to note that the unaided HEIs covered in this study are the undergraduate colleges that offer courses in liberal arts and are located very close to the vicinity of the students. Thus, students (particularly from marginalized sections of the society) access these colleges with low cost. But, students enrolled in government and aided HEIs usually stay in the hostels or private accommodations (as these institutions are largely located in district headquarters) and spend more on non-fee items. Also, majority of the students accessing these institutions belong to relatively well-off families (within lower social groups) and therefore spend more on their education.

Several studies reveal that the probability to enrol and continue in higher education increases with the availability and accessibility of financial assistance to students, as it helps them to spend more (Glocker 2011; Iriti et al. 2017). Furthermore, it helps better to the students belonging to poor and marginalized sections of the society as they are not in a position to cover their costs for higher education. The regression results show that the amount of the scholarship received by the students (lnPMSS_amount) is positively related to the annual per capita total household expenditure on higher education and the household spending on non-fee items. The regression coefficient of lnPMSS_amount for the first household expenditure function (lnEducost1) finds that If the amount of scholarship is increased by 1%, we expect total household spending on HE to increase by 1%, and the expenditure on non-fee items of HE will go up by 2%.

Conclusion

This chapter examines the variability of household expenditure on higher education in rural Odisha, one of the backward states of India, using the data collected through a student survey in two tribal dominated districts of Odisha (Mayurbhanj and Keonjhar) in 2016–2017. The findings suggest that the annual average household expenditure on higher education among marginalized sections of the society in rural Odisha is around 30% of the annual family income. Of the total household expenditure on HE, students have spent only about 5% on fees and the rest on non-fee items such as private tuition, food and accommodation, transport, Internet, etc. Interestingly, the non-fee expenses on higher education by the households vary significantly between the students residing in hostels and day scholars. Households have spent close to one-fifths of their total expenditure on higher education annually on private tuition or private coaching. It is observed from the interaction with the students that majority of them go for private tuition as the quality of teaching in the colleges is very poor. As the students belong to socially backward communities and are mostly first generation learners, they need more support to perform better in class. Fee paid for private tuition or coaching is relatively high for science streams, followed by commerce and then arts. While discussing about this with the institution heads, shortage of teachers came as a serious issue to provide quality higher education to the students as many HEIs appoint temporary/guest teachers to teach. However, a detail discussion on this issue is outside the scope of this chapter, though it is an important issue and needs urgent attention both in academics and policy domain.

Students enrolled in government and aided HEIs have spent more than the students enrolled in unaided higher education institutions and interestingly, this difference is largely due to the difference in the payment of non-fee items. The regression results suggest that the ‘type of institution’ had a significant effect on household expenditure on higher education. Students enrolled in government and aided institutions have incurred higher level of expenditure than the students studying in unaided institutions and this is true in both total and non-fee household expenses. This suggests that though students access low cost higher education in unaided colleges (as these are located close to vicinity), quality is a serious concern. Many of these undergraduate colleges in rural Odisha are running with very few teachers and seriously lack physical infrastructure that were evident from the fieldwork.

The household expenditure on higher education in rural Odisha varies widely across socioeconomic groups such as gender, caste and family income. The results presented show pro-male bias in household spending on higher education, i.e. the household expenditure on higher education is more for male students than for female students in rural Odisha. The pro-male bias in household spending is little more in case of non-fee items than the total household expenditure on higher education. The analysis of the annual per capita household expenditure on higher education by caste shows that OBC students spend more than SC and ST students in both fee and non-fee items. Thus, ST students spend the least on their higher education and it may be due to their poor economic status. As expected, students belonging to poor households have invested less on higher education than the households with better income. Thus, from a policy perspective, reducing the financial burden of poor households in rural areas is the need of the hour to provide wider access to HE among them – a target that is being set up since independence and emphasized consistently thereafter. This may be possible by allocating more public funds on need-based scholarships to the students belonging to the marginalized sections of the society, particularly in rural areas.

When considering and interpreting the findings of this chapter, some limitations should be borne in mind. First, the willingness of the household to investment on higher education is shaped by their choice for institutions and courses of study and thus, it could be argued that a discussion on parental preferences for HEIs and relating it with household expenditure would have been better. Second, the data is collected from students who are currently pursuing undergraduate and postgraduate courses of select disciplines, and therefore it may not provide the complete picture of household investment on HE in rural Odisha (e.g. the expenses on costly disciplines like engineering and management are not taken into account). Third, an interaction with parents and getting their experiences on managing the expenses on HE of their wards should have been done to provide a better picture on the issues discussed in this chapter. Despite these caveats, this chapter makes a contribution to the economics of education literature by unfolding the story of household investment in higher education among marginalized sections of the society in the context of rural Odisha – an area which is grossly ignored both in academia and policy domain. The analysis in this chapter should be useful to policymakers seeking to design interventions aimed at increasing participation in higher education in rural Odisha, particularly among marginalized sections of the society.

Appendix

Table 10.5 Institution-wise number of students surveyed for the study

Table 10.6 Sample students by type of institution, gender, caste, religion and region

Table 10.7 Notation and definition of the variables used in the regression analysis

Table 10.8 Summary statistics of the variables used in the regression analysis