I am back in Lucknow since a week or so, and this time concentrate on economic questions - both those relating to real estate (see here and here) and to the scandal of poverty. Both issues are frequently linked, of course, since many people claim that Muslims could not find housing in new Lucknow because they are by and large poor (see my earlier discussion of residential segregation). As a quantifiable basis to discuss this proposition, I had analysed data from the Public Distribution System (PDS) before my departure. The key finding: Muslims in Lucknow are poor - but not poorer than their non-Muslim counterparts.

Despite its devastating findings on Muslim poverty, the Sachar Committee Report - most quoted in this context - emphasized wide variation within the heterogeneous Muslim community, something that is often forgotten: that most Muslims in India are poorer than non-Muslims need not imply that this is also true in the context of Lucknow. Most existing micro-level studies that explore Muslim poverty in the city in turn focus exclusively on the Muslim poor rather than assessing their predicament relative to the whole population - a serious design flaw in my opinion. Given that better sources are unavailable, and considering that I do have neither time nor resources for a proper independent survey, this post thus uses a rather unusual data source: the Public Distribution System. My analysis of Muslim poverty is based on the complete lists of Antyodya, BPL and APL ration card holders in Lucknow, whom I further classified as Muslim or non-Muslim based on their names.

Though without practical alternative, this source is not unproblematic of course: the PDS is notorious for inefficiency and corruption. A first problem is the fact that most of the information in the PDS database is self-reported, and people are likely to overstate the extent of their deprivation and under-report their income in a bid to gain access to the system (this problem is arguably more pronounced among richer beneficiaries). Once one gained access to the system, however, there is no incentive to under-report further and neither is there a reason to assume that the extent of under-reporting differs between Muslims and non-Muslims. In consequence, the income distributions can still be meaningfully compared even if absolute figures are dubious.

A second problem with PDS data is the heavy inflation with bogus beneficiaries: over 620000 ration cards were issued to households in urban Lucknow alone, despite the fact that even the 2011 Census counted only 577510 households in the city, only a fraction of which are likely to qualify for ration cards. In hindsight, this gross inflation however also makes it more likely that a potential under-representation of Muslims stems from the disproportionate inclusion of bogus non-Muslim cardholders rather than from Muslims' deliberate exclusion. Corrupt officials have an incentive to maximize rents by adding fake beneficiaries and selling off the subsidized rations allocated to these on the open market; the exclusion of Muslims in contrast has no such economic benefits but would carry considerable electoral risks, especially in a city like Lucknow where the Muslim minority is both sizeable and vocal. If these assumptions hold, one can however change the baseline and compare the absolute number of Muslim and non-Muslim card holders against the (uninflated) 2001 Census.

Based on such a comparison, I estimated Muslim and non-Muslim poverty for various parts of Lucknow district. For the core city, the comparison suggests a maximum head count poverty rate of at most 2.8% among Muslims (lower to the extent that bogus beneficiaries carry Muslim names and higher to the extent of Muslims' deliberate exclusion from the PDS), which is almost the same as the non-Muslim rate of 3.0%. Both figures might at first seem extremely low compared to state-wide poverty estimates of around 30%, but this can be explained with the narrow definition of "urban" Lucknow in the PDS, which excludes most of the shanty towns and slums on the city's outskirts. Indeed, a look at rural parts of Lucknow district reveals a much higher head count poverty rate of 41.7% for Muslims and 39.0% for non-Muslims; in small towns, they rise further to 41.7% for Muslims and a staggering 67.1% for non-Muslims. Since it remains unclear how exactly the urban boundary in the PDS differs from that used for the Census, the most meaningful rates might in the end be those for the whole district: 11.9% for Muslims and 16.7% for non-Muslims. Like other Muslims across India, many Muslims in Lucknow are thus poor - but so are many non-Muslims.

The comparison of Muslim and non-Muslim income distributions comes to similar conclusions; unlike head count rates, these remain unaffected by the number of bogus beneficiaries and/or Muslims' deliberate exclusion. The average annual income of Muslim households in both the BPL and the Antyodya categories across the district is 15309 Rs, which does not differ much from the 14742 Rs earned by poor non-Muslim households; average annual per capita income stands at 4884 Rs for Muslims and 5239 Rs for non-Muslims, again no major difference. Even if we allow for some distortion in the PDS, these figures thus hardly support the claim of exceptionally high Muslim poverty either.

The above histogram finally shows the annual household income distribution for all PDS beneficiaries, including those above the poverty line. The non-Muslim distribution is more spread out than that for Muslims, but again, the differences however remain marginal, in fact so much so that a Kolmogorov-Smirnov test finds them to be statistically insignificant. Let me repeat this once more: the differences between Muslim and non-Muslim household income in the lower economic strata (that is the sectino of society covered by the PDS) are so small as to be statistically insignificant. In other words: they could as well be random.

In conclusion, data from the PDS thus show that many Muslims in Lucknow are poor - more so than non-Muslims if seen on the basis of income per capita, less so in head count rates and household income - but also that the figures differ so little between both groups that they can hardly account, for example, for a rather stark pattern of residential segregation. Moreover, the household income distribution of those Muslims above the poverty line but still within the ambit of the PDS closely resembles that of non-Muslims, the extent of property ownership is similar in both groups, and the ratio between poor and non-poor beneficiaries actually suggests relatively more Muslim poor in new Lucknow than in the old city. None of these statistical findings supports the "poverty hypothesis" that Muslims were excluded from new Lucknow because of limited financial means, even if one allows for considerable distortion in raw data.

I will be back soon (I hope) with news on how poverty is discussed in Lucknow's Muslim circles - but first, I have to go for one of those delicious kebabs and meet a couple of friends. If you have any comments meanwhile, fire away below!