in the report.

T

Muslims’ relatively poor socio-economic status and the steps needed to improve things. In this commentary I explore some of the methods used in the report, and I argue that they tell us far less than we might wish about the processes of discrimination and the major factors that may be responsible for Muslim disadvantage, which prevents us from designing the best government and NGO interventions to deal with problems the minority community faces. I then make some suggestions about the kinds of analyses that might be done in the future.

The committee has collected a great deal of information from government departments, the census, private survey organisations such as NCAER as well as from a few smaller surveys it conducted itself. The report’s tables and technical notes

Economic and Political Weekly March 10, 2007

alone run to around 130 pages. However most of this information, extremely valuable though it is to have it updated and all in one document, does not really change our fundamental picture of Muslims’ socioeconomic progress so much as it adds some additional quantitative support to conclusions already reached by others on the basis of studies and research over the past three decades. The committee finds that Muslims are generally worse off than most other communities in terms of their access to public and private sector jobs, education, infrastructure and credit, and that the gap between Muslims and other communities has failed to close or has even increased on some dimensions over the past few decades. The committee also finds that, while there is still a significant fertility difference between Muslims and Hindus overall (a fact the committee handles very gingerly), large numbers of Muslims do use family planning and the gap in fertility differences between the minority and majority communities is closing (SCR: 40-42). None of these findings, though, would come as a surprise to people familiar with the annual reports of the Minorities Commission in the decade after 1978 or with the various data on the relative socio-economic standing of Muslims and their representation in public and private employment compiled by Syed Shahabuddin when he edited the journal Muslim India. As far back as 1991, for instance, Asha Krishnakumar published an article on Muslim demography and fertility rates in Muslim India that makes many of the same points as the Sachar Committee about the community’s growth rate.2 Nor would the data on the economic concentration of Muslims in short-term contractual work or self-employed occupations and in handicrafts come as a shock to people familiar with the excellent surveys done by NCAER, or the NSS data reported in the EPW in the 1990s by one of the Sachar Committee’s members.3 Much of this data was conveniently gathered together, with a great deal of analysis and history on Muslims’ experiences in post-independence India besides, in an excellent book written by Mushirul Hasan in 1997.4

Old Data in New Bottles?

So in terms of data on the extent of Muslims’ under-representation in various fields, access to services and credit, and overall levels of poverty there is not much new here, although having said that it is clear that the Sachar Committee Report has been much more widely read and discussed (as well as condemned) than any of these previous reports, which gives it considerable value regardless of novelty. Neither, unfortunately, does the report tell us as much as we might hope about just why Muslims should be worse off, and what India can do about it. Are the causes of Muslim poverty specific to Muslims, in which case a community-specific strategies such as reservations might be appropriate for some sections of the community at least, or are they better addressed by general anti-poverty or educational programmes? The problem here, as the authors acknowledge at various times in the report, is that most of the data collected by the committee are not really suitable for helping us determine whether Muslims are doing badly because of their economic backwardness, levels of Muslim political inclusion or exclusion, anti-minority discrimination, cultural factors that may influence factors such as the community’s access to education or healthcare, or some other state-specific or community-specific factor.

The data used in the report are largely of two kinds. First, we have opinions on the extent and causes of Muslims’ socioeconomic backwardness, put forward by various people who either voluntarily contacted the committee or else were asked to present to it. Many of the complaints and theories about why Muslims are worse off – for instance, claims that Muslims have difficulty in getting admission for their children in good schools, or claims about widespread discrimination in employment or housing markets – are detailed in chapter 2 of the report, accompanied by a cautious caveat that these opinions are not necessarily shared by the committee. The second kind of data that the committee uses are outcome measures such as, for example, the proportion of Muslims compared to members of other socio-religious communities (SRCs), who are employed in various occupations, who have received credit from banks, or who have participated in various government programmes.

The first kind of data shed a lot of light on Muslim perceptions of why their community has poor representation in employment and education, has high levels of poverty, and poor access to housing and services. Complaints about discrimination in various spheres of life loom large, and many have the ring of truth about them, not least because they repeat claims made in many other publications over the years.5 But these are complaints from a largely self-selecting group of individuals and organisations, rather than representative opinion polls or scientifically designed studies that allow us to truly compare Muslim experiences with the experiences of members of other Indian communities. As a result, the complaints detailed in chapter 2 can be easily dismissed by those on the right as being either unrepresentative or without foundation. For instance, we have plausible allegations from Muslims that “Discriminatory practices, especially at the time of the interview, were…reasons for poor Muslim representation even at the class IV level or in grade D employment where high educational qualifications are not required” [SCR: 20]. But assessing the validity of such claims in order to inform public policy is difficult for two reasons. First, we know that similar claims are also made by members of other communities, such as the OBCs and SCs, who do much better than Muslims in obtaining public employment. And second, without systematic data on the numbers of people from various SRCs applying for class IV or grade D positions, it is very difficult in practice to assess just how much worse Muslims are doing than other groups.

In fact, the committee was able to obtain data on number of applicants, interviews, and rates of acceptance for all communities only for the very highest positions in higher education (admission to the IIMs) and the civil services (the results of the UPSC competition). And, somewhat surprisingly, and in an improvement from at least one earlier study, it found that Muslim candidates in fact had success rates for these exams that were the same as other applicants in the case of the UPSC and slightly higher than average in the case of the IIMs (SCR:68, 165-166). Now the IIM and UPSC selection processes are, of course, widely regarded as perhaps among the fairest application processes in the country, so finding no great evidence of discrimination there does not necessarily tell us about whether there is discrimination against minorities in other spheres of employment. But the general point remains that we need this kind of data about applicants and their qualifications as well as their success rates if we are to make inferences about whether discrimination might exist. The many individual tables in the SCR showing us the degree of Muslim under-representation in government

Economic and Political Weekly March 10, 2007 service or higher education cannot tell us whether this under-representation was because of the lack of qualified Muslim candidates, community preferences for some occupations over others, overt discrimination against Muslims, the greater political and social influence of other groups in society, or any one of a dozen other plausible explanations.

Interventions and Their Limitations

And in the longer term, unless we can get a good sense of the relative importance of these multiple potential causes of Muslim disadvantage, it will be difficult to design meaningful public policy interventions that will truly attack the roots of Muslim poverty and disadvantage. Instead, we will have no choice but to simply list a variety of possible solutions, without knowing how significant any one of these might be. While this will be attractive to some politicians – because it leaves them maximum autonomy to adopt or not to adopt particular policies while still nonetheless claiming that they are “doing something” that the committee has recommended they do for the community – it will not help us to identify the most cost-effective and valuable policies to help Muslims and others who need help.

What would a good research strategy be, one that builds on the excellent datagathering work of the Sachar Committee but also moves us further in the direction of identifying which of the causes of Muslim disadvantage might be the most important? First, we need to acknowledge that the bivariate analysis of the kind used in most of the tables and chapters in the report

– in which we are separately shown in various chapters and tables the relationship between being Muslim and outcome A (poverty), then outcome B (education), then outcome C (fertility), then outcome D (access to credit) – is much less illuminating than analyses which show us the relationship between being Muslim and, say, education, controlling for most of the other factors that might also plausibly influence an individual’s education, such as her gender, income, caste, level of parental education, etc. Statistics cannot explain everything, to be sure, and not everything that is important for explaining Muslims’ economic disadvantage can be measured statistically, but multivariate analysis is surely essential if we want to weigh the relative contribution of different factors to Muslim poverty, as well as work out whether Muslim backwardness is really due to Muslim-specific issues or is rather the general outcome of economic and educational disadvantage.

A limited amount of this kind of statistical analysis, to be fair, is carried out in the report (SCR: 72-74, 107-08). And fascinating stuff it is too, though for some reason we are only told the broad direction of the results, and not given the regressions themselves, which would help us assess their validity. In the first regression the authors use data on 20-30 year olds from the NSSO 61st round to find out “if SRC status affects educational attainment even after we control for economic status and other factors” (SCR: 72-73). After controlling for age, gender, economic status, state of residence and rural/urban residence, they find out that being Muslim (and being SC/ ST) has an important independent effect on reducing an individual’s chance of graduating from college, especially for men and in towns and cities (SCR: 73-74). In the second statistical analysis, the authors try to find out whether, controlling for a variety of attributes, Muslims have lower work participation rates than other communities. The report finds some small differences but emphasises that “it needs to be noted that lower probabilities for Muslims finding work in urban areas is not necessarily a reflection of discrimination... the estimates are based on data generated after job choices have been made. If a person chose not to apply/join a regular job, s/he has made a voluntary choice” (SCR: 107-08). The author then identifies greater policy initiatives in education, credit and skill acquisition as the way to go, because these seem to have the greatest prospect of improving things for Muslims, especially the relatively large proportion of Muslims who are self-employed.

These few pages provide some of the highest quality analysis and best recommendations in the whole report – if you read only one page I would make it p108

– and one wishes that whoever wrote these sections these had been given a freer rein to shape the data collection and analysis in the report as a whole. But that said, the presentation of the statistical results could be made even more helpful. For example, in the analysis of the reasons for different graduation rates (pp 72-74), as well as being told that Muslim identity has an effect on reducing the prospects of graduation, we are also told that “As expected, the results show that economic status has a very large, positive, and significant impact on Graduate Attainment Rates (GARs)”. The obvious question from a policy perspective is whether the effects of being Muslim outweigh those of being poor, or male, or living in a town in explaining graduation rates. Calculating such substantive effects is relatively easy and it would be very informative to be told, for instance, what would happen to Muslim graduation rates in Kerala or Uttar Pradesh if we increased average income or number of average years in school or some other variable by a certain percentage. Seeing the effects of specific policy interventions presented in this way would make the extent of the different benefits and trade-offs much more transparent for politicians and the public, which would help focus the public debate.

One model of what might be done with the wealth of data the Sachar Committee has collected is provided by an excellent recent paper on earnings inequality in India by Sumon Kumar Bhaumik and Manisha Chakrabarty, a paper that like Sachar uses NSS data.6 Bhaumik and Chakrabarty use a variety of statistical analyses to establish the key role of educational attainment in explaining earnings differentials, and they also establish that the returns to education for Muslims compared to equally qualified Hindus have been declining in relative terms over the period from 1987 to 1999, a period when inter-caste differences among Hindus have declined. This generates at least two hypotheses: one is that general overt discrimination against Muslims is the cause (although this is brought into question by their finding that returns to education for Muslims in 1987 were slightly higher, unless we think discrimination has increased); the second and more likely hypothesis is that reservations for Muslim backward classes (MBCs) and OBC Hindus introduced over the past 25 years may provide these Hindu communities with better educational endowments and a wider range of good job options, at least in public service, compared to Muslims.7

To further improve the statistical analysis using the Sachar Committee’s data we also need to introduce a wider range of statistical controls. The committee at various times highlights the importance of state-level characteristics (Muslims in some states consistently do better than in others), and of patterns of political access and incorporation, for instance, something I have dealt with in my own work, but these variables are not then brought into the analyses in the report as explanatory factors for Muslim disadvantage, perhaps because they are seen as politically sensitive.8

Economic and Political Weekly March 10, 2007

Another variable that needs to be controlled for is the estimated proportion of illegal migrants from Bangladesh into India. If we are to believe the estimates inadvertently provided by minister of state for home Prakash Jaiswal to Parliament in mid-2004, there are over 12 million illegal immigrants from Bangladesh in the country, the majority of whom are Muslims and who no doubt declare themselves as Indian in the census and in the social surveys from which the data on Indian Muslims used in the Sachar Committee are drawn.9 These migrants are almost certainly poorer and less well educated and have less secure employment on the whole than Indian Muslims, and their presence in the sample is likely therefore to depress our estimates of how well Indian-born and educated Muslims might be doing compared to their SC/ST, OBC and upper caste Hindu counterparts.

How much of a difference might adjusting for migrants make? Let us assume for the sake of argument that there are 13 million illegal migrants from Bangladesh in India, a figure somewhat higher than the estimate presented by Prakash Jaiswal in mid-2004 but several millions lower than the 15 million estimate given by the previous BJP government. If we estimate the Muslim proportion at 80 per cent of this total (given the fact that Hindus emigrate disproportionately from Bangladesh), that would give us perhaps 10 million Bangladeshi Muslims in India, or 8 per cent of the total Indian Muslim population. According to the Sachar report (appendix table 4.1), Indian Muslims have a 59.1 per cent literacy rate, which places them well below the national average of 65.1 per cent, with a literacy rate even below that of SCs/ STs. If, for the sake of argument, however, we were to assign the 2002 Bangladesh literacy rates – 31 per cent for women and 50 per cent for men – to 8 per cent of this total, then the literacy average for the remaining Indian Muslims would rise by almost a full per cent nationally, and by somewhat more than this in states such as Assam and West Bengal in which illegal migration has been the greatest. On other factors such as our estimates of Muslim poverty levels the difference might potentially be even greater.10

More Research Needed

The second step that needs to be taken, after the multivariate statistical analysis of the Sachar Committee data, is to conduct specially commissioned surveys and research studies to find out if some of the most likely hypotheses about the reasons for Muslims’ poor representation in many areas of Indian life are indeed correct. In representative opinion polls, do Muslims report facing greater levels of discrimination in seeking housing, jobs or employment than other SRCs? Can we identify research strategies to find out if empirically this is the case, and if so how important discrimination is in explaining outcomes compared to other factors such as poor access to good schools, credit, or skills training?

Such studies have been used with increasing success in the study of discrimination in other societies, and they are especially useful where, as perhaps is the case in India, many members of the majority need some convincing about the challenges faced by minorities. In the US, for instance, years of studies showing that African Americans were doing worse than whites on many dimensions, and were less likely to be employed in particular occupations than whites did not necessarily convince the majority that discrimination was at work. But more recent studies have effectively established that much discrimination is taking place. Economists Marianne Bertrand and Sendhil Mullainathan, for instance, had fictitious but equally qualified applicants with recognisably African American and white names (Lakisha and Jamal vs Emily and Greg) apply for 1,300 jobs, and were able to establish that the African American applicants got 50 per cent fewer callbacks for interviews than applicants with recognisably white names.11 Other studies, while acknowledging the continuing role of prejudice against minorities in the US, have found that a large proportion of the difference in black/white success in the labour market is caused not by prejudice but by differences in skill levels between these groups, which suggests a renewed focus on improving African American access to education and skill acquisition at the elementary and high school level, where too many from the community remain in sub-standard educational environments.12

After reading this report, I kept thinking about two factors that, whichever way the data were analysed, seem to be crucial in explaining how well Muslims fare: access to education, and the particular state in which Muslims live. Much of the analysis in Sachar focuses on education, but the report still leaves the reader without a good sense of which factors leave Muslim children, and especially Muslim girls, without sufficient access to primary, secondary and higher education. The main explanation put forward by the committee seems to be generalised discrimination and lack of access, in part because of poverty. But this cannot be the total answer, because in some states Muslim participation in education is much higher than the norm. In fact, on many measures Muslims are doing better in western and southern states than they are in the rest of the country. Just to give one example, although in general Muslim literacy rates are below the Indian average, in 10 of the 21 states studied (including Maharashtra, Gujarat, Andhra Pradesh and Karnataka) Muslim literacy rates are actually higher than average (SCR: 53). So there is clearly something that is state- or region-specific that seems to be making a major difference to Muslims’ life chances. This raises two issues for the future. One is whether we can identify the factors that, in these states, seem to be responsible for Muslims having a greater degree of access to education. The second, and clearly more intractable issue concerns India’s poorest and least well-governed states, through which centrally sponsored schemes have to operate. Can India design appropriate policies to help its most economically disadvantaged people, including its minorities, even if the state machinery through which it has to work in some states is not up to the task?

EPW

Email: swilkinson@uchicago.edu