Gender and public spending on education in Pakistan: a case study of disaggregated benefit incidence.

Sabir, Muhammad

To what extent has government education spending in Pakistan been effective in reducing gender gaps in enrolments? To answer this question, this article reviews the benefit incidence of government education spending. It finds that government subsidies directed towards primary education are pro-poor in all four provinces of Pakistan, but females are at a disadvantage in terms of access to primary education. Government subsidies directed towards higher education are poorly targeted, and the poorest income group receives less than the rich income group; indeed the policies favour those who are better off. Similarly, gender disparity in terms of access to public subsidy is higher at the tertiary level, and the lowest at the primary level, which also reflects poor targeting. Improving access of the poor as well as better female participation involve not simply rearranging the public subsidies but also addressing the constraints that prevent the poor and females from accessing these services.

1. INTRODUCTION

It is generally believed that education is one of the basic rights of every human being, irrespective of sex, age, creed, religion, etc. Moreover, the target of universal primary education cannot be achieved without female access to educational opportunities, which contains several external benefits. In addition, access to educational opportunities assumes prime importance for empowerment of women. However, inequalities in access to education between males and females can be found in many countries across the world including Pakistan. According to conventional wisdom, a combination of cultural, social, and economic factors are responsible for placing young girls and women at a serious disadvantage vis-a-vis access to school and the prospect of completing their education. This disadvantage can be altered through public policies including gender sensitive public spending on education.

The above assertion about the role of public policy is based on the theory of public finance, (1) which demonstrates that public expenditure on education can affect the population in a number of ways, which has significant gender dimensions. For example, government spending on primary education is likely to generate more income for women than spending on universities, for the simple reason that there are relatively more women primary school teachers than women university lecturers. Moreover, these expenditures provide subsidised educational services, which is a form of "in kind transfers". These "in-kind transfers" improve the current well-being of the recipients, and enhance their longer-run income-earning potential. They can be considered as both current and capital transfers to the recipients, and therefore can be termed as the "benefit incidence" of public spending.

The main concern of this paper is to assess the gender dimension of the "benefit incidence". The study has two basic objectives. First and foremost, it aims to investigate which income group actually benefits from the government's subsidised education services? Second, how are these benefits distributed between males and females? In other words, this article is an attempt to capture the gender dimension of public spending on education in Pakistan through benefit incidence analysis, which may help to better understand the root causes of gender inequality in Pakistan and provide a guideline for developing a gender sensitised education policy.

There are three reasons for analysing gender inequality in public spending on education in Pakistan. First, it is one of the most important services that could empower the Pakistani women of today. Second, education spending, especially at the primary level, is considered to produce positive external benefits, and therefore, a strong case can be made for the continued involvement of the government for gender equitable public spending on education. Finally, in 1990s the government of Pakistan initiated a project named Pakistan Integrated Household Surveys (PIHS), which consisted of four round running through 1995-96 to 1998-99 under Social Action Plan (SAP). The objective of PIHS, a national sample survey, was to provide household and community level data, which could be used to monitor, evaluate, and assess the impact of SAP. The PIHS data provides information on the income of households and gender disaggregated enrolments in public and private schools, colleges and universities, which offer an opportunity to estimate the distribution of government subsidies in the education sector at provincial levels.

The paper begins, in Section 2, with a brief review of the benefit incidence approach and establishes how gender dis-aggregations can be readily incorporated in the analysis. Section 3 presents the result of Benefit Incidence of educational spending in Pakistan at provincial level. Section 4 highlights the regional gender inequality by using data from Pakistan Integrated Household Survey (PIHS) 1998-99. Section 5 makes some concluding observations and offers some policy implications.

2. THE 'BENEFIT INCIDENCE' OF PUBLIC SPENDING

The technique employed in this paper to assess gender differentials in public service provision is 'benefit incidence analysis'. This has become an established approach in estimating the distribution of public expenditures since the path-breaking work on Malaysia by Meerman (1979) and on Colombia by Selowsky (1979). (2) There has been a recent resurgence of interest and incorporation of gender dimension in the approach, reviewed in Van de Walle and Nead (1995); Van de Walle (1998) and Demery (2000).

The technique usually involves a three-step methodology. First, estimates are obtained on the unit cost of providing a particular service. These are usually based on officially reported public spending on the service in question. Second, these unit costs are then imputed to households, which are identified (usually through a household expenditure survey) as users of the service. Households, which use a subsidised public service in effect, gain an in-kind transfer, which depends on the unit subsidy involved (say the subsidy per primary school enrollment) and the number of units consumed by the household (the number of children currently enrolled in a public primary school). Finally, aggregated estimates of benefit incidence are obtained in groups ordered by income. In brief, benefit incidence analysis measures the distribution of in kind transfers across the households.

Expenditure incidence analysis therefore brings together two sources of information. First, data on the government subsidy (estimated as the unit cost of providing the service less any cost recovery back to the government) allocated to the education. Second, information on the use of public education services by individuals and households, which is usually obtained from household surveys.

The disaggregated benefit incidence analysis by gender based on the assumption that government provide in-kind subsidy by providing subsidised schools, colleges and universities and to gain the subsidy, households must enroll children at publicly subsidised educational institutions. If households typically send more boys than girls to these publicly-funded schools, there will be a gender difference in benefit incidence, simply because more of the government subsidy will be utilise boys than girls.

As earlier mention, the three steps for disaggregated benefit analysis can easily be transformed mathematically by considering the group-specific benefit incidence of government spending on education:

[X.sub.j] [equivalent to] [3.summation over (i = 1)] [E.sub.ij] [S.sub.i]/[E.sub.i] [equivalent to] [3.summation over (i = 1)] [E.sub.ij]/[E.sub.i] [S.sub.i] .... .... .... ... ... (1)

[X.sub.j] is the value of the total education subsidy imputed to group j. [E.sub.ij] represents the number of school enrolments of group j at education level i, and [E.sub.i] the total number of enrolments (across all groups) at that level. [S.sub.i] is government net spending on education level i (with fees and other cost recovery netted out), and i (=1,..,3) denotes the level of education (primary, secondary, and tertiary). The share of the total education subsidy (S) accruing to the female ([x.sub.j]) is given by:

[X.sub.j] [equivalent to] [3.summation over (i = 1)] [E.sub.ij]/[E.sub.i]([S.sub.i]/S) [equivalent to] [3.summation over (i = 1)] [e.sub.ij][s.sub.i] .... .... .... ... ... (2)

Clearly, this share is determined by two factors: the share of the gender in total enrolments at each level of education ([e.sub.ij]), and the share of each level of education in total education spending (s;). [e.sub.ij] is determined by household enrolment decisions, whereas [s.sub.i] reflects government spending allocations.

3. EDUCATION SUBSIDIES IN PAKISTAN

Under the constitution of Pakistan, education is a provincial subject. Consequently, provincial governments are primarily responsible for financing the provision of educational services. However, federal government plays an important role in the financing of higher education. This financing is administrated by University Grants Commission (UGC), (3) which provides financial support to general and technical universities, colleges and institutes. As a result, calculation of unit subsides for higher education contains both provincial and federal public expenditure.

3.1. Public Education System in Pakistan

The structure of Pakistan's public education system has the following main characteristics described as follows: first, there is the basic education. The basic education system consists of early childhood education (or pre-primary schooling), which is optional for children 3 to 5 years old and called kachi Pehli. After that primary education where the official entry age is 5 year and ideally should be completed in 5 years at the age of 9 years. The next level, following primary education, is secondary level education, consists five years of education after primary education started from the age 10 years and ideally ended at the age of 14 years.

Subsequently level of education after secondary education is tertiary level with two options available to students who may choose polytechnic Institutes and colleges for technical education, and general colleges/schools for higher secondary education. Finally, after completion of two-year higher secondary education next level of education encompasses three lines of study: a system of technological/engineering colleges and universities, medical colleges and universities, and general colleges and universities.

However, from the perspective of public finance there are four broad categories, which generally reported in budget documents of provincial and federal governments. These are primary, secondary, general colleges and universities, and finally technical and professional institutes, colleges and universities. In this report, we used all four categories for the analysis of incidence of public spending in education and named than primary, secondary, tertiary (included general colleges and universities) and professional education (included technical institutes, professional colleges and universities).

3.2. Source of Data

The information on the income of households and enrolments in public schools, colleges and universities at various levels of education is taken from the micro data of Pakistan Integrated Household Survey (PIHS) Round 3: 1998-99. PIHS is a national survey conducted by Federal Bureau of Statistics, which provides household and community level data on various indicators related to education, health, water sanitation and population welfare. The data on public spending on education is taken from the federal and provincial demand for grants and appropriation 1999-2000. Finally the data on cost recovery is taken from Estimates of Receipts 1999-2000 for each respective province.

3.3. Unit Subsidies in Education

Table 1 presents the result of province-wise estimates of unit subsidies in education. Second column of Table 1 represents the gross unit subsidy--current cost to the governments of a student study in a particular level in a public institution. It is calculated as total recurrent spending of provincial and federal governments on a specific level of public education divided by total number of students of same level in' the province. Third column corresponds to any revenue from cost recovery from a specific level divided by number of students of same level. Finally, the last column represents the net subsidy for all levels and for all provinces, which is simply the difference of Column 2 and 3.

Table 1 reveals the regional disparities in the unit subsidies in education, which also varies with level of education. For instance, in case of primary education, the amount of unit subsidies is highest in Sindh, following Punjab and NWFP, lowest in Balochistan. However, this does not imply that Sindh spends the highest amount among the four provinces, because, it is the combination of two factors: demand of public education and total public expenditures on a particular levels.

Unit cost estimates reported in Table 1 are limited in several respects. First, due to unavailability of actual public spending on education, estimates of unit subsidies are based on provincial and federal revised estimates of public spending on education. According to the historical trend in public finances of the country, actual spending on education may appear less than revised estimates. As a result, these reported estimates of subsidies may contain an upward bias and overly stated amount of subsidies. Second, in the absence of disaggregated data, unit subsidies were obtained as provincial averages, ignoring gender and urban rural variations. It may lead to over or under estimation of regional and gender unit subsidies. Furthermore, number of students for the estimation of subsidies have taken from PIHS data after multiplying with blow up factor and may contain sampling or non sampling errors which over or under estimates the amount unit subsidies.

4. GENDER DISPARITY IN EDUCATION

There are several ways to measure gender differentials in education. Gross and net enrolment rates often reveal gender differences, especially when reported by income quartiles, similarly completion and drop out rates are another way to highlight gender disparities in education. However, gross enrolment is the widely used basic indicator, which highlights gender disparity at the first stage of the education and indicates lake of access to educational facilities for women. Therefore, as a first step, gender disparity in education is highlighted by gross enrolment ratio and subsequently gender disaggregated estimates of benefit incidence are presented.

4.1. Gender Disparity in Gross Enrolment Ratio

Table 2 presents the province-wise gross enrolment rates (4) divided into three broad categories; primary, secondary and tertiary, and by income quartiles, and gender. The gross enrolment rates show typical biases in enrolment behaviour, with males being more likely to be enrolled in school, and with the bias becoming more noticeable with higher levels of schooling. Another interesting point, which emerges from Table 2 is provincial variation in gender specific enrolments. Gender disparity in education is higher in Balochistan and NWFP as compared to Sindh and Punjab.

4.2. Gender Disaggregated Benefit Incidence Estimates

By combining the unit cost of the public education system with the use of public schooling facilities by household, we can estimate the benefit incidence of government spending on education. The province-wise

results of this exercise (based on the subsidy schedule of Table 1) are reported in Table 3A, 3B, 3C and 3D respectively. Three types of disaggregation are reported: first, the subsidy is distributed across the four income quartiles starting from the richest 25 percent to the poorest 25 percent of the population; second, the benefit incidence estimates are disaggregated by education level; and finally, it is reported by gender.

(a) Punjab

Three clear messages emerge from the estimates of benefit incidence (see Table 3A). The first message is that education spending is reasonably progressive at the primary level; the subsidy to the poorest quartile forms a higher share than the subsidy to the richest quartile. This progressiveness is particularly striking in relative terms; the poorest 25 percent population of Punjab received more than 35 percent of the subsidy. While, the richest 25 percent receive only 13 percent of the subsidy at the primary level (see the last column of Table 3A). However, in the case of subsidy at the secondary level the main beneficiaries are lower and middle-income groups, which indicates that subsidy is less progressive. In contrast, educational subsidy at tertiary level is regressive and the highest share received by the richest 25 percent, which receive more than 50 percent of the subsidy at the tertiary level.

The second message is that the progressiveness at the primary level is illusionary because the share in subsidy it is not adjusted by the group wise population. The per capita estimates of level-wise subsidy presents a real picture of the incidence. According to this indicator the lowest per capita subsidy received by the poorest 25 percent of the population in all three educational categories. However, this picture varies with the level of education, it is relatively better at the primary level and worse at the tertiary level.

Finally, the gender disparity persists at all educational levels and in all income classes. It clearly emerges, that the pattern of subsidy is biased towards females, for instance, at the primary level 45 percent, at the secondary level 39 percent and at the tertiary level 45 percent share of the subsidy spent on females and rest on males. This result is also consistent with the per-capita estimates and in all cases per-capita estimates are lower for females and higher for males. In addition, females belonging to the poorest 25 percent population receive the lowest share in education subsidy at all levels of education and it is lowest at the tertiary level.

(b) Sindh

Similar to Punjab, estimates of benefit incidence portray the same picture in Sindh (see Table 3B). Education spending is reasonably progressive at the primary level; the subsidy to the poorest quartile amounts to a higher share than the subsidy to other income groups and the share of, the subsidy decreases gradually with increase in income. The poorest 25 percent population of Sindh receives more than 35 percent of the subsidy. In contrast, the richest 25 percent receives only 11 percent of the subsidy at the primary level (see the last column of Table 3B). A similar pattern exists at secondary level with a slight change in magnitude ranging from 31 percent to the poorest income group and 14 percent to the richest income group. However, this pattern reverts at the tertiary level, the main beneficiaries are the richest income group, which receives 35 percent of the subsidy. The subsidy at the tertiary level of education is regressive in nature and share of the subsidy decreases with decline in income level and the lowest share in the subsidy at tertiary level received by the poorest income group, which is-only 16 percent.

Similar to Punjab, per-capita estimates of the subsidy is not consistent with the share-wise benefit incidence and lowest per-capita subsidy at all educational levels is received by the poorest income group. However, the highest per capita subsidy is received by the poorest 25 percent of population at the primary level and lowest at the tertiary level of education.

The relative disadvantage of females with respect to access to education follows a steady pattern in Sindh. It is lowest at the primary level, where they receive 41 percent of the total subsidy for primary education; this disadvantage gradually increases with the level of education and is greatest at the tertiary level, where they receive only 21 percent of the total subsidy at tertiary level education. This pattern confirms the hypothesis that relative disadvantage increases with the level of education. Similarly, in per capita terms, females receive lower subsidy in each income group at all educational levels.

(e) NWFP

In contrast to Punjab and Sindh, estimates of benefit incidence demonstrate a different pattern in the NWFP (see Table 3C). Education spending is not reasonably progressive at any level of education. Moreover, except primary education, poorest population receives the lowest share in educational subsidies; particularly at the tertiary level the poorest income group receives only 8 percent of the subsidy. A similar pattern exists also at the secondary level with a slight change in magnitudes ranging from 22 percent to the poorest income group and 28 percent to the upper middle-income group. However, the pattern of subsidy distribution is slightly different at the primary level, where the main beneficiaries are the low middle-income group, which receives 33 percent of the subsidy, and 28 percent is received by the poorest income group, and 24 percent is received by the upper middle income and 15 percent is received by the richest income group.

As in the case of the other provinces, per-capita estimates of the subsidy are inconsistent with the share of subsidy and the lowest per-capita subsidy at all levels of education receive by to the poorest income group. However, the highest per capita subsidy is received by the poorest 25 percent of population at the primary level and lowest at the tertiary level of education.

The relative disadvantage to females was lowest at the primary level, where they receive 37 percent of the total primary subsidy and the greatest at the tertiary level, where they receive only 29 percent of the total subsidy for the tertiary level. It is also interesting to note that the relative disadvantage of females is inversely correlated with level of income. For instance, females in the poorest quartile of income receive only one-third of the primary education subsidy received by the poorest 25 percent and males receive remaining two-third. In contrast, females in the richest quartile of income receive almost half of the primary education subsidy received by the richest 25 percent and males receive remaining half. This disadvantage further increases with the level of education--females in the poorest quartile receive only 5 percent of the secondary education subsidy and almost zero percent at tertiary level.

(d) Balochistan

as in the case of NWFP, estimates of benefit incidence demonstrate that education spending is not reasonably progressive at any level of education in Balochistan (see Table 3D). Moreover, except primary education, poorest population receives the lowest share in educational subsidies; particularly at the tertiary level of education; the poorest income group receives only 8 percent of the subsidy. A similar pattern exists at the secondary level with slight change in magnitudes ranging from 21 percent to the poorest income group and 33 percent to the upper middle-income group. However, this pattern differs at the primary level, where the main beneficiaries are low middle-income group, which receive 30 percent of the subsidy, and 26 percent is received by the upper middle income group, and 23 percent receive by the poorest income group, and 21 percent by the richest income group.

Similar to other provinces, per-capita estimates of the subsidy are not consistent with the share and lowest per-capita subsidy at all levels of education and the lowest per capita subsidies are received by the poorest income group. Moreover, this picture varies with the level of education, relatively better at primary level and worse at the tertiary level.

In Balochistan, the pattern of relative disadvantage to females with respect to access to education is similar to that in other provinces. However, magnitudes of subsidies present a bleaker picture. It is the lowest at the tertiary level, where they receive only 11 percent of the total subsidy for professional education and the greatest at the primary level, where they receive 34 percent of the total primary subsidy. Similarly females receive lowest per capita subsidies in each income group at all levels.

5. REGIONAL GENDER DISPARITY

Gender disaggregated benefit incidence results can be used to provide a comparative picture of regional gender disparity. To grasp the comparative picture of regional gender disparity, we computed female-male ratio of benefit incidence of public spending by region. The result of this exercise is presented in Figure 1. It is clear from Figure 1, that the relative disadvantage of females to access public education was higher in rural areas as compare to urban areas. All rural areas and urban Sindh and Balochistan follow the same pattern of disadvantage--higher disadvantage with higher level of education.

The disadvantage is the greatest in rural Balochistan and Sindh. This observation is compatible with the expectation that female enrolment rates decrease with the level of education. In contrast, it emerges from Figure 1, that there is relative disadvantage to males with respect to access to public education in urban Punjab at the primary and the tertiary levels, and urban NWFP at tertiary level.

[FIGURE 1 OMITTED]

6. CONCLUDING OBSERVATIONS

Gender disaggregated benefit analysis of public spending on education can be used to describe why the poorest income quartile has less access to public education specially at the higher level of education. The answer is to be found, in part, in the greater gender enrolment bias among the poorest sections of the society. For example in Balochistan, males in the poorest quartile receive 14 percent and 17 percent of the subsidy at primary and secondary level, while females in the poorest quartile receive 9 and 4 percent of the subsidy at the primary and secondary level respectively. However, females in the richest income group receive 9 and 8 percent of the subsidy at the primary and secondary level respectively, which is share-wise equal at primary level and higher at secondary level with compare to females of the poorest segment of the society. This trend also persists in the secondary education subsidy of other provinces excluding Sindh and tertiary and professional level subsidies of all provinces where females of poorest gain less than rich groups. The gender inequality, therefore, is a critical component of overall inequality in the benefit incidence of education spending.

These results powerfully demonstrate how public spending on education benefit males more than females in Pakistan in 1998-99, and this in turn influences the overall inequality of education spending in the country. It does so for a combination of reasons: first, households choose to enrol males more than females at all levels of schooling (see Table 2), therefore, a gender bias would be present in the benefit incidence of public spending on education; second, public spending is not properly targeted to the regions of higher gender disparities and as a result, public spending is not sufficiently contributing in the reduction of gender disparity.

Based on the benefit incidence analysis three sets of policies can be recommended to improve gender equality in the society. First, the poorest segment of the society receives the lowest per capita subsidy; therefore, public policies related to public spending on education should be targeted towards the region with higher level of poverty. Second, in the presence of higher regional gender inequality, region specific education policy may be helpful for the gender equality, specially public spending in rural areas on female education will play a vital role as compare to urban areas. Finally province-wise policies related to gender equality in education at various level works better than the national policies. For instance, in Balochistan, a reasonable proportion of the government budget is devoted to schooling services which females tended not to use--tertiary education institutions. A shift of spending towards primary and secondary schooling would lead to an improvement in the share of the total budget going to females (as well as to poorer groups in the community). In contrast, such policy is not helpful for the other regions particularly in Punjab and NWFP where female enrolments were higher at tertiary public institutions. Therefore, a shift of spending towards tertiary level would lead to an improvement in the share of the total budget going to females. However, such decisions should not rest on benefit incidence estimates alone. They should also be based on a sound understanding of how household behaviour would be affected by such expenditure switches.

REFERENCES

Balochistan, Government of (1999-2000) Demands for Grants and Appropriations, Current Expenditures. Finance Department.

Balochistan, Government of (1999-2000) Estimates of Receipts for 1999-2000. Finance Department.

Demery, Lionel (1996) Gender and Public Social Spending: Disaggregating Benefit Incidence. Poverty and Social Policy Department, World Bank (Unprocessed).

Demery, Lionel (2000) Benefit Incidence: A Practitioner's Guide. Poverty and Social Development Group Africa Region. The World Bank.

Meerman, Jacob (1979) Public Expenditures in Malaysia: Who Benefits and Why? New York: Oxford University Press.

NWFP, Government of (1999-2000) Demands for Grants, Current Expenditures. Finance Department.

NWFP, Government of (1999-2000) Estimates of Receipts for 1999-2000. Finance Department.

Pakistan, Government of (1998-99) Pakistan Integrated Household Survey 1998-99. Federal Bureau of Statistics, Islamabad.

Pakistan, Government of (1999-2000) Details of Demand for Grants and Appropriations 1999-2000, Current Expenditure, Vol. 1. Finance Division, Islamabad.

Punjab, Government of (1999-2000) Estimates of Charged Expenditures and Demands for Grants, Current Expenditures. Finance Department.

Punjab, Government of (1999-2000) Estimates of Receipts for 1999-2000. Finance Department.

Selowsky, Marcelo (1979) Who Benefits from Government Expenditure? New York: Oxford University Press.

Sindh, Government of (1999-2000) Estimates of Charged Expenditures and Demands for Grants, Current Expenditures. Finance Department.

Sindh, Government of (1999-2000) Estimates of Receipts for 1999-2000. Finance Department.

Van de Walle, Dominique (1995) The Distribution of Subsidies Through Public Health Services in Indonesia, 1978-87. Chapter 9 in Van de Walle and Nead.

Van de Walle, Dominique (1998) Assessing the Welfare Impacts of Public Spending. World Development 26:3, 365-379.

Van de Walle, Dominique, and Kimberly Nead (1995) Public Spending and the Poor: Theory and Evidence. Washington, D. C.: Johns Hopkins and World Bank.

Van de Walle, Dominique, and Kimberly Nead (1995) Public Spending and the Poor: Theory and Evidence. Baltimore, Md: Johns Hopkins University Press.

Comments

The case study on gender and public spending on education in Pakistan will be a timely and much-needed addition to the growing body of documentation on this vital issue. It becomes all the more important in view of two recent publications, and a third publication due to be launched tomorrow. The two reports in hand are the Pakistan Human Condition Report (2002, CRPRID, Planning Commission), and the Pakistan Poverty Assessment Report (2002, World Bank). The forthcoming document is the Mahbub ul Haq Human Development Centre's annual Human Development Report for South Asia.

While the findings of the study under review do not come as a major surprise to anyone working on either education or gender issues, it is necessary to have the evidence at hand, to be able to convince both the policy-makers and the donors. To that end, this study will be useful.

But therein lies one of its constraints--it does not go the extra mile to translate its findings into policy-level recommendations--or even strategic observations and conclusions. It just states the findings, and leaves it at that. One must draw one's own conclusions as well as any policy prescriptions.

The assessment of gender inequality through "benefit incidence" analysis of public expenditure on education is a useful mechanism, but the introductory rationale for promoting female education is still the same old utilitarian and functionalist approach to women's empowerment issues--once again ignoring the rights-based approach, despite the token first sentence.

On the data source, just a brief comment--the Government, especially the Ministries of Finance and Economic Affairs and Planning and Development, have recently been very critical of the PIHS, particularly in the context of the World Bank's poverty assessment exercise. Thus, the author may perhaps wish to review his data source, or look for more acceptable alternative sources.

In a more equal world, we would not see the assertion on p. 1, whereby we note the ghettoisation of women in low-income employment such as primary school teachers vs. university professors, where the author appears to be promoting the former as an income generating activity for women. While we commend the promotion of primary education and its gender dimension, we also note with dismay that the author does not refer to female "employment", but instead to "income generation" for women. This is, once again, a reflection of the traditional marginalisation of women's work vis-a-vis that of men in male perception.

In the explanation on unit subsidies in education (p. 6) it is stated that because education is a provincial subject, the provincial governments are responsible for financing educational services. The author omitted mentioning the fact that tertiary (university) education is federally financed--through the former University Grants Commission (UGC), recently renamed the Higher Education Commission (HEC). Hence, provincial expenditure statements do not provide a complete picture of public expenditures on education at all levels.

On pp. 8-9, gender differentials are discussed only in regard to enrolment rates. However, as we are all aware, enrolment rates are just one indicator it is necessary to also provide gender- and urban/rural-disaggregated data pertaining to the primary level completion and dropout rates. We know that the dropout rates are much higher for girls, especially in the rural areas--being as high as 90+ percent in some parts of Balochistan and NWFP. More analysis is required on this unconscionable aspect of gender disparities in education, particularly at the primary level.

The findings on the contradictory aspects of increasing subsidies on female primary education vs. continuing gender disparities in enrolment--at all educational and income levels--needed some discussion, analysis and recommendations. This is a major omission in the paper.

Also missing is any mention of the availability of educational institutions in the non-government (NGOs) and private (for-profit) sectors. While this may not appear to be relevant to the discussion on government expenditures or subsidies on education in the formal public sector, it does have an impact on national gross enrolment rates at all levels--primary, secondary and tertiary--both in the general and technical/professional streams, especially in the urban areas. The existence of the NGOs and the private sector cannot be ignored. The issue of quality---or the lack thereof--has not been introduced in the discussion, despite its importance.

Whether by commission or omission, the paper received for comments four days before the conference, is incomplete, as per the outline given on p. 3 (incidentally, it would be helpful for discussants if the pages were to be numbered). Only Punjab data is provided, and the paper ends abruptly at the start of the discussion on Sindh, while NWFP and Balochistan are missing altogether, as are the promised observations and policy implications. The triple-spaced 11-page paper comprises introductory statements, definitions and three tables on school enrolments and subsidies. Thus, it emerges that the author's basic question in the abstract: "To what extent has government education spending in Pakistan been effective in reducing gender gaps in enrolments?" has not been answered in the current form of the paper.

Tahira Abdullah

Islamabad.

Author's Note: The author gratefully acknowledges comments and suggestions by Hafiz A. Pasha, Director, Regional Bureau for Asia and Pacific, UNDP, Haroon Jamal, Principal Economist, SPDC, and the conference participants, specialty Tahira Abduilah and Khadija Haq.

(1) See Demery (2000).

(2) See Van de Walle (1998).

(3) Recently, University Grants Commission was renamed as the Higher Education Commission.

(4) Both public and private enrolments are included.

Muhammad Sabir is an Economist based at Social Policy and Development Centre (SPDC), Karachi.

Table 1

Education Unit Subsidies in 1998-99

(Rs/Anum)

 Education Unit Subsidies
 (Per Student)

 Cost
 Gross Recovery Net
Primary Education
 Punjab 2,686 30.20 2,656
 Sindh 3,100 1.62 3,098
 NWFP 2,201 0.11 2,200
 Balochistan 1,555 -- 1,555

Secondary Education
 Punjab 1,445 79.07 1,366
 Sindh 2,093 63.23 2,029
 NWFP 2,915 71.79 2,843
 Balochistan 2,605 2.44 2,603

General and Professional Colleges/
Universities/Institutes
 Punjab 5,538 562 4,976
 Sindh 3,655 299 3,356
 NWFP 9,172 200 8,972
 Balochistan 7,126 53 7,073

Source: Author's estimates based on Provincial and Federal Demand for
Grants 1999-2000, Estimates of Receipts 1999-2000, and PIHS 1998-99.

Table 2

Gross Enrolment Ratio by Gender, Quartile, and Province (1998-99)

 Primary Secondary
Income Level /Province Males Females Males Females

Punjab
 Rich 101% 81% 89% 74%
 Upper Middle 92% 86% 78% 62%
 Lower Middle 82% 72% 59% 45%
 Poor 65% 53% 39% 25%
 All Group 80% 68% 62% 46%
Sindh
 Rich 87% 73% 105% 59%
 Upper Middle 80% 55% 82% 51%
 Lower Middle 66% 47% 61% 38%
 Poor 49% 35% 47% 29%
 All Group 64% 46% 65% 40%
NWFP
 Rich 105% 80% 106% 56%
 Upper Middle 88% 55% 79% 36%
 Lower Middle 80% 56% 61% 24%
 Poor 67% 31% 53% 11%
 All Group 81% 50% 70% 28%
Balochistan
 Rich 80% 67% 76% 47%
 Upper Middle 78% 33% 76% 26%
 Lower Middle 68% 43% 50% 14%
 Poor 54% 36% 51% 13%
 All Group 69% 43% 62% 22%

 Tertiary
Income Level /Province Males Females

Punjab
 Rich 16% 12%
 Upper Middle 8% 5%
 Lower Middle 3% 3%
 Poor 2% 1%
 All Group 7% 5%
Sindh
 Rich 25% 10%
 Upper Middle 16% 6%
 Lower Middle 12% 4%
 Poor 9% 4%
 All Group 15% 6%
NWFP
 Rich 17% 10%
 Upper Middle 11% 5%
 Lower Middle 7% 1%
 Poor 3% 0%
 All Group 9% 4%
Balochistan
 Rich 22% 4%
 Upper Middle 12% 2%
 Lower Middle 5% 1%
 Poor 4% 0%
 All Group 11% 2%

Source: Estimates based on PIHS 1998-99.

Table 3A

Benefit Incidence of Public Spending on Tertiary Education-Punjab

 Male

 Per Capita Share Per Capita
Income Level (Rs) (%) (Rs)
 Primary Level
 Rich 1,544 8% 1,012
 Upper Middle 1,536 11% 1,419
 Lower Middle 1,593 16% 1,448
 Poor 1,341 20% 1,065
 All Group 1,475 55% 1,233
Secondary Level
 Rich 824 14% 681
 Upper Middle 852 17% 594
 Lower Middle 680 16% 497
 Poor 468 15% 249
 All Group 670 61% 455
Tertiary Level
 Rich 529 28% 435
 Upper Middle 261 17% 183
 Lower Middle 84 5% 94
 Poor 74 4% 45
 All Group 229 55% 179

 Female
 Total
 Share Per Capita Share
Income Level (%) (Rs) (%)
 Primary Level
 Rich 5% 1,292 13%
 Upper Middle 10% 1,478 21%
 Lower Middle 15% 1,521 30%
 Poor 15% 1,206 35%
 All Group 45% 1,356 100%
Secondary Level
 Rich 9% 760 22%
 Upper Middle 11% 729 27%
 Lower Middle 12% 590 28%
 Poor 8% 359 22%
 All Group 39% 566 100%
Tertiary Level
 Rich 24% 482 53%
 Upper Middle 12% 222 29%
 Lower Middle 6% 89 12%
 Poor 3% 59 7%
 All Group 45% 203 100%

Source: Estimates based on PINS 1998-99 and Provincial Demand for
Grants 1999-2000.

Table 3B

Benefit Incidence of Public Spending on Tertiary Education-Sindh

 Male Female Total

 Per Per Per
 Capita Share Capita Share Capita Share
Income Level (Rs) (%) (Rs) (%) (Rs) (%)

Primary Level
 Rich 1,410 6% 1,133 5% 1,276 11%
 Upper Middle 1,658 13% 1,162 9% 1,413 22%
 Lower Middle 1,537 18% 1,015 12% 1,273 31%
 Poor 1,236 21% 907 15% 1,075 36%
 All Group 1,422 59% 1,011 41% 1,218 100%
Secondary Level
 Rich 1,294 9% 664 5% 972 14%
 Upper Middle 1,263 17% 766 10% 1,025 27%
 Lower Middle 1,028 18% 638 10% 839 28%
 Poor 830 20% 523 11% 684 31%
 All Group 1,034 64% 626 36% 837 100%
Tertiary Level
 Rich 682 29% 177 6% 452 35%
 Upper Middle 393 22% 135 6% 276 28%
 Lower Middle 284 16% 84 4% 192 20%
 Poor 201 11% 87 5% 145 16%
 All Group 371 79% 115 21% 252 100%

Source: Estimates based on PIHS 1998-99 and Provincial Demand for
Grants 1999-2000.

Table 3C

Benefit Incidence of Public Spending on Tertiary Education-NWFP

 Male Female Total

 Per Per Per
 Capita Share Capita Share Capita Share
Income Level (Rs) (%) (Rs) (%) (Rs) (%)

Primary Level
 Rich 1,436 8% 1,223 7% 1,325 15%
 Upper Middle 1,578 15% 1,039 10% 1,318 24%
 Lower Middle 1,629 22% 1,128 12% 1,410 33%
 Poor 1,329 19% 646 9% 983 28%
 All Group 1,492 63% 945 37% 1,228 100%
Secondary Level
 Rich 2,129 14% 1,129 8% 1,600 23%
 Upper Middle 1,882 19% 917 9% 1,420 28%
 Lower Middle 1,602 19% 675 8% 1,139 27%
 Poor 1,398 18% 303 5% 805 22%
 All Group 1,694 70% 679 30% 1,173 100%
Tertiary Level
 Rich 934 29% 616 19% 773 48%
 Upper Middle 732 23% 223 8% 470 31%
 Lower Middle 333 11% 54 2% 184 13%
 Poor 214 8% 5 0% 111 8%
 All Group 536 71% 209 29% 368 100%

Source: Estimates based on PIHS 1998-99 and Provincial Demand for
Grants 1999-2000.

Table 3D

Benefit Incidence of Public Spending on Tertiary Education--Balochistan

 Male Female Total

 Per Per Per
Income Level Capita Share Capita Share Capita Share
 (Rs) (%) (Rs) (%) (Rs) (%)
Primary Level
 Rich 1,091 12% 972 9% 1,038 21%
 Upper Middle 1,158 20% 477 6% 858 26%
 Lower Middle 1,028 20% 667 11% 865 30%
 Poor 820 14% 551 9% 690 23%
 All Group 1,019 66% 636 34% 844 100%
Secondary Level
 Rich 1,774 16% 1,021 8% 1,416 24%
 Upper Middle 1,939 28% 636 5% 1,450 33%
 Lower Middle 1,266 17% 363 5% 824 22%
 Poor 1,311 17% 348 4% 831 21%
 All Group 1,563 77% 539 23% 1,091 100%
Tertiary Level
 Rich 1453 44% 199 6% 851 50%
 Upper Middle 576 25% 78 3% 337 28%
 Lower Middle 369 12% 78 2% 232 15%
 Poor 223 8% 6 0% 119 8%
 All Group 631 89% 87 11% 372 100%

Source: Estimates based on PINS 1998-99 and Provincial Demand for
Grants 1999-2000.