Security Research Review

A World Bank-funded study titled “Religious School Enrollment in Pakistan : A Look at the Data” attempts to challenge this conventional wisdom, arguing that the problem of madrassas has been overstated. Authored by Tahir Andrabi, Jishnu Das, Asim Ijaz Khwaja and Tristan Zajonc, the study argues that most analysts have inflated the number of madrassa students in Pakistan and that madrassa recruitment has been broadly stable rather than rising. Using data from the 1998 census, three rounds of the Pakistan Integrated Household Survey (1991, 1998, 2001) and a sample educational survey in three districts of Punjab, the authors argue that notwithstanding claims of madrassa enrollment of up to 1.5 million students, their data shows that a maximum of 475,000 students in Pakistan are enrolled in madrassas, and that the actual number is likely lower (less than 200,000). This means that no more than 1% of school-going Pakistani children aged 5-19 attend madrassas and that “there are 38 times as many children in private and 104 times as many in government schools compared to madrassas”. The authors also argue that, apart from small bumps in 1998 and 2001 related to regional political events, there has been no sustained increase in madrassa enrollment. 

These findings have not gone unchallenged. Samina Ahmed, who authored a 2002 International Crisis Group study of madrassas, has questioned both the sources of data and discrepancies between the World Bank study and official Pakistani statistics. Pointing out that the controversial 1998 census was flawed and that the household survey is not designed to elicit information about madrassa enrollment, Ahmed says that “ Ejazul Haq , Pakistan ’s Federal Minister of Religious Affairs, says that the country’s madrassas impart religious education to 1,000,000 children”. She states that the claim that madrassa enrollment has been constant is contradicted by the Ministry of Education’s 2003 directory of madrassas that lists 10,430 madrassas, up from 6,996 in 2001. Ahmed uses this finding to question the study’s claim of constant madrassa enrollment.

How might we reconcile these very different views? One discrepancy is simply definitional. Andrabi et al. consciously focus on full-time madrassa enrollment, excluding cases where a student attended a madrassa part-time for evening classes on the Quran and so on, whereas the Pakistani government includes part-time enrollment in its definition. This exclusion appears reasonable if one holds the theory that madrassas are problematic because students are educated exclusively with a curriculum based on religion with little thought given to secular subjects. A student attending a madrassa part-time is likely to have been exposed to alternative modes of thinking and is better able to adapt to the modern, secular world. But arguably even part-time exposure to radical ideology could be destabilizing. To complicate matters further, a small number of radical madrassas (such as Binori town in Karachi and Dar-ul Uloom Haqqania Akora Khattak in the North West Frontier Province, NWFP) has produced the largest number of militants, which would suggest that overall religiosity as reflected by the growth of madrassas is only weakly correlated with radicalism.

Rather than an overall measure of madrassa enrollment, it might then be better to focus on a subset of Deobandi or Wahhabi madrassas that are affiliated to radical organizations (although this suggestion admittedly takes us beyond the scope of the present review). A large number of militants have arisen from the middle class, be it Al Qaeda, Laskhkar-e-Tayyiba or Lashkar-e-Jhangvi, who were not simply produced by madrassa education but by political events (e.g. war in Afghanistan ) and state connivance. Even the rise of the fundamentalist Muttahida Majlis-e-Amal (MMA) alliance in NWFP and Balochistan was a product of army policy, and the MMA is believed to have lost support in recentl months despite the widespread presence of madrassas. A large madrassa-educated population is more likely to pose a problem for Pakistani social policy than for state survival, particularly because violent activism is the product of jihadist organizations rather than radicalization per se. In short, the authors’ definitional decision is reasonable given the theory of how madrassas feed radicalism, although the theory itself is limited as an explanation of radicalism in Pakistan or elsewhere.

But a more serious issue with the study is that of potential rural bias. The 1998 census was controversial because it is said to have intentionally undercounted the population of Karachi in an attempt to minimize the claims of mohajirs, Urdu-speaking migrants from India who in the 1980s mobilized into a political force under the leadership of the present-day Muttahida Qaumi Mahaz (MQM). If the census undercounted the population of Karachi , it is also likely to have minimized madrassa enrollment in a city that accounts for 10% of the country’s population. To their credit, the authors use the household survey and data from a their own survey of villages in 3 (of 103) Punjabi districts as a corrective, revising their estimates upwards, but the political interference in the 1998 census does render it a questionable source.

The study does contain some interesting insights and suggestions for future research despite these question marks. For one, the study is a rare attempt to use different and unusual sources of data to build a composite picture. The analysis finds that (1) with the exception of a predictable concentration in the Pashtun belt bordering Afghanistan and Karachi, madrassa enrollment is thinly spread across the country and not concentrated in specific districts (say, in central Punjab). The authors also argue that (2) Pashtun culture does not explain the prevalence of madrassas in the Pashtun belt since their own district survey finds no greater propensity among Pashto speakers to send their children to madrassas (although they do not provide the raw data to allow a reader to judge whether any finding could be statistically significant).

Contrary to the widespread picture of under-invested government schools giving way to madrassa education, the study finds that (3) non-religious private schools are the most dynamic part of the education sector. The authors also argue that (4) there is no evidence for a dramatic increase in madrassa enrollment, and that any discernable increases appear linked, unsurprisingly, to the 1979 Soviet invasion of Afghanistan (together with Zia-ul Haq’s ascension to power) and the 1998 rise of the Taliban. While Pakistani government figures appear to contradict this finding, the hypotheses that madrassa enrollment is linked to geopolitical events is at least plausible.

Finally, the authors identify some interesting variations in the propensity of households to send children to madrassas using data from their survey of 150,000 households in villages in three districts (although they take pains to clarify that these are not definitive statements). Somewhat surprisingly, the authors argue that (5) households defined as “radically Islamic” are not more likely to enroll children in madrassas, although their proxy variable is whether a household names at least one child Osama (or some variation of that name). This is a rather sketchy operationalization and better data is undoubtedly required. They also find that (6) the share of private schools increases as households become wealthier, but that the share of madrassas does not change. The authors also provide some further findings on intra-household variations in madrassa enrollment but a closer examination of these patterns lies outside the scope of this review.

In short, the World Bank-funded study offers some fascinating insights and uses an innovative methodology, but the possibility of rural bias leaves a question mark over its findings. The paper’s major contribution is to provide an independent empirical study of the phenomenon and to push available data in a counterintuitive direction. The results will hopefully encourage other social scientists to bring their own analysis to the issue, and the ensuing debate could in the end produce a consensus view.