A Closer Look At How Partition Changed Punjab’s Religious Map

When I wrote my first post on this blog about a year and a half ago, I only had access to religious data at a district-by-district level for Punjab (and also Bengal). When I wrote my post on Kashmir, I switched to using sub-district (or tehsil) data, which is more detailed. Since then, all of my posts on Partition have used that more detailed data from the 1931 Census. I decided to revise the Punjab map using the tehsil data. I won’t rewrite the whole post here though I encourage you to read it here. Instead I’ll compare the original map to the new one, and discuss a few features that the increased resolution reveals.

Apart from the obvious improvement in quality, and my addition of much of the North-West Frontier Province to the second map, there are a few details worth commenting on, though obviously the overall picture remains largely unchanged. First there was an overwhelmingly Muslim pocket due south of Delhi, far from any other Muslim majority area. Second, the increased detail really emphasizes the concentration of Sikhs in the southern half of what is now the Indian state of Punjab. The northern half of Indian Punjab was very mixed, with large Hindu, Muslim, and Sikh populations. Third, Amritsar and Tarn Taran tehsils, which are the Sikh-plurality (blue) tehsils directly in the middle of the province, were mostly surrounded by Muslim-majority or Muslim-plurality tehsils. This will be relevant when we look more closely at the partition lines later in the post. Finally, I noticed the Sikh-heavy area west of Amritsar centered on Lyallpur (modern-day Faisalabad). This part of Punjab was the focus of a major British plan to irrigate previously arid parts of Punjab and turn them into agricultural centers. Often, Sikhs migrated to these so-called canal colonies, where they were given plots of land to farm. The west-central part of Punjab, where most of these colonies were, tended to be Muslim-majority, but with a larger Sikh than Hindu population. Farther west, the Muslim majority became more pronounced, but the minority population was mostly Hindu. Below is a map of the non-Muslim Punjab population, to reveal the Hindu/Sikh distribution (the “Sikh” category actually includes all other religions, but the Sikhs made up the overwhelming majority of the non-Muslim, non-Hindu population).

I adjusted the opacity so that the colors look more faded as the non-Muslim share of the population decreases. As you can see, the Sikhs are mostly concentrated in the central Punjab. Hindus are much more numerous than Sikhs in the south and west of the state. I think this may be due to the canal colonies, which were located in the west-central Punjab and were a major magnet for Sikh migrants in the early 1900s. Farther west and south, the main minority is presumably the ten or fifteen percent of the original Hindu population who didn’t convert to Islam during the seven centuries or so of Muslim rule over western Punjab.

In this post, I experimented with a few different ways of presenting the data. The map below, requested by Vikram, a commenter, is a map depicting the Muslim vs. non-Muslim breakdown, which was the basis of the British partition lines.

Now here is the same map, but with the partition line drawn in.

Here we see clearly that the Radcliffe Line (the official name of the border between India and Pakistan) in Punjab was very favorable to India, as I noted in my original post. Not a single non-Muslim-majority tehsil or princely state ended up in Pakistan, while nine Muslim-majority tehsils or states ended up in India. Furthermore, seven of these were contiguous or near-contiguous with Pakistan. Here is the same map with the nine Muslim-majority tehsils given to India highlighted.

The seven Muslim tehsils near the Pakistan border plus the two non-Muslim tehsils they surround, Tarn Taran and Amritsar, had an overall religious profile of 52.0 percent Muslim, 27.3 percent Sikh, 18.4 percent Hindu, and 2.1 percent Christian. Therefore, theoretically, the concept of  “two nations” would have been better served if the whole block had gone to Pakistan. That way, only four tehsils would have been “stranded” on the wrong side of the line, two per country. That still wouldn’t have been a satisfying solution to me though because the goal of Partition should have been to minimize displacement of people, not create a solution that was abstractly “fair” to both India and Pakistan. I think a better option would have been to give as much of this Muslim-majority block as possible to Pakistan while keeping Tarn Taran and Amritsar in India. There would have been two obvious options for achieving this goal. One would have been for Kapurthala state, southeast of Tarn Taran and Amritsar to accede to India while the remainder of the Muslim tehsils went to Pakistan. Alternatively, Gurdaspur and Batala tehsils, just to the north of Amritsar, could have remained with India while the other Muslim-majority tehsils and Kapurthala state went to Pakistan.

I judged which of these would have been better by assuming that all “stranded” Sikhs and Hindus would have moved to India, all Muslims would have moved to Pakistan, and Christians would have stayed put, as that is how Partition played out in general. Then I calculated how many refugees would have been expected under each scenario. The first scenario, Partition as it actually occurred, resulted in about 1,359,000 refugees from this pocket of Muslim-majority areas given to India. The second scenario, Kapurthala staying with India, would have generated 1,167,000 refugees and would have looked like this:

The third would have involved Gurdaspur and Batala staying with India, would have resulted in 1,177,000 refugees, and would look like this:

A possible fourth option would have been for the whole area to go to Pakistan. That would have resulted in 1,195,000 refugees and look like this:

Obviously, if we reject the rules set up by the British and think outside the box, there are other possible outcomes that could have reduced or eliminated refugees. Some such possibilities include a multi-religious neutral zone in central Punjab, an independent and united Punjab, or no Partition of India at all. However, accepting the rules as defined by the British, the second option of Kapurthala going to India would have had the fewest refugees. Unsurprisingly, the British chose the one that created the most refugees. It is quite possible that the British wanted to make sure that the Sikhs, who were left in a very bad situation by Partition, would not be forced to abandon Amritsar, their holy city. Thus all the Muslim tehsils around Amritsar were given to India to make Amritsar more defensible and less exposed. It isn’t obvious to me though that the border in option two above was dramatically less defensible. Amritsar was more exposed, but any border through that part of Punjab would be difficult to defend given the flat topography. And the British left Pakistan divided in two and separated by the length of India, presumably a much greater obstacle to defensibility than a zigzag in the Punjab border. It should also be noted that the actual border drawn in that exact region proved impossible to defend for Pakistan in the 1965 war, when India invaded and ended up on the outskirts of Lahore by the time a cease-fire was declared.

Perhaps the British were willing to give India a favorable deal in Punjab (and as commenters have pointed out, in Bengal too) as a way of convincing the Congress leadership to accept Pakistan’s creation. I think that the main reason the British supported Pakistan’s creation is that it kept the most strategically valuable parts of British India in somewhat friendly hands. The North-West Frontier Province, parts of Kashmir (which the British expected to go to Pakistan), and perhaps Balochistan were strategic in the Cold War with the Soviet Union, which was already underway by 1947. As long as those territories ended up in Pakistan, the British probably weren’t too picky about what else ended up where. They may have promised and delivered India a favorable deal to ensure that India’s leadership wouldn’t get difficult and prevent the British from leaving the subcontinent quickly and unobtrusively. The British also wanted to ensure that India’s leadership wouldn’t drag its heals on recognizing Pakistan’s legitimacy (India voted in favor of Pakistan’s admission to the UN, Afghanistan was the only country opposed). So maybe the generous amount of land given to India in Punjab and the strategic headwaters in Bengal were meant to quietly grease the wheels for the British Empire’s withdrawal from South Asia and Pakistan’s entrance onto the world stage.

Finally the original goal of this project was to examine how Partition affected the religious demographics of Punjab, so I will post the original map with the 1931 populations, and the map with 2001 (India) and 1998 (Pakistan) populations so you can see how Partition changed Punjab forever. It is especially worth noting the change in the Muslim-majority teshils that remained in India, which must have lost a majority of their population in 1947, and the narrow Muslim-majority tehsils in Pakistan, which lost just a bit less than half.

How Did Partition Affect North India?

Last time I wrote here, I looked at Sindh, a province that was overwhelmingly Muslim in 1931, but had a few regions with Hindu majorities. I wondered why Sindh hadn’t been partitioned along with Punjab, and I got some really great responses from readers in the comment section. I have a theory now that I think makes sense, which I will get to later in this post. First though, I want to discuss north India. As I mentioned in the Sindh post, a large group of Urdu-speaking Muslims from north India migrated to Sindh upon Partition. I decided that an examination of how they changed their new homeland was incomplete without looking at how their departure affected the place they left. Did north Indian Muslims leave in large enough numbers to have an impact on the overall demographic profile of north India? The Muhajirs, as the migrants and their descendants are known, didn’t all come from one place, and typically any person who voluntarily moved to Pakistan in 1947 is defined as a Muhajir. However, the largest number came from the Urdu-speaking north, mainly the state of Uttar Pradesh (then called the United Provinces), and to a lesser extent Bihar. In this post, I chose to focus on those two states, as well as the modern-day states of Uttarakhand and Jharkhand, which were separated from Uttar Pradesh and Bihar respectively only a few years ago.

Like Sindh, these provinces of British India were not partitioned in 1947; instead India got all of Bihar and the United Provinces of Agra and Oudh. The Muslim population of both provinces was relatively sizable though, and had traditionally wielded significant power. The Mughal capitals of Agra and Fatehpur Sikri are in UP, as is Lucknow, which was the cultural capital of north India in the post-Mughal era and is still one of the centers of Shia Islam in South Asia. Deoband and Bareilly, the founding locations of the Deobandi and Barelvi movements of Sunni Islam, are both in Uttar Pradesh. UP was also an early source of strength for the Muslim League. In the 1937 elections, the Muslim League won 26 seats in the United Provinces, compared to one seat in the North West Frontier Province, Punjab, and Sindh combined (these territories make up most of modern day Pakistan).

As usual, I’ll give a sketch of the basic demographic picture pre-Partition. According to the 1931 Census, Bihar, which included the modern-day states of Bihar and Jharkhand (as well as Odisha, which I excluded here), was 80 percent Hindu, 12.7 percent Muslim, and 7.3 percent Other. In this case, most of the “Others” were followers of the so-called “tribal” religions that may predate Hinduism in South Asia. There was also a fairly sizable Christian minority, numbering in the several hundred thousands. The United Provinces had broadly similar demographics: 84.4 percent of its population was Hindu, 15 percent Muslim, and 0.7 percent Other. Obviously, the big difference between Bihar and UP was the presence of the tribal religions in Bihar.

Both Bihar and the United Provinces were classified in the 1931 Census as overwhelmingly Hindustani-speaking. Hindi and Urdu, which are two standardized forms of Hindustani, are in fact spoken in Uttar Pradesh. This classification is a bit problematic for Bihar, because today the language spoken in Bihar is typically classified either as a separate language (Bihari), or as a cluster of related languages. Linguistically, the most interesting part of the region is the Chota Nagpur division, which is roughly equivalent to today’s Jharkhand state, which at the time was in southern Bihar. The 1931 Census suggests that 47.5 percent of the population was Hindustani-speaking. Obviously, we don’t know how those languages would be classified today, but according to Wikipedia, the Bihari languages spoken in Jharkhand include Maithili, Khortha, and Angika. I don’t know how accurate that is, but if correct, it suggests that Chota Nagpur division’s “Hindustani” speakers in 1931 spoke a Bihari dialect and not modern Hindi.

The remaining 52.5 percent of Jharkhand/Chota Nagpur’s population is even more interesting. 21 percent were Bengali speakers, concentrated near the border with Bengal. These areas were transferred to West Bengal after Partition. 3.4 percent spoke Oriya, which is the dominant language of neighboring Odisha (then called Orissa). 6.6 percent spoke Kurukh, a Dravidian (South Indian) language. 18.8 percent spoke one of three Munda languages – Santali, Mundari, and Ho. These languages are really unusual, as they belong to the Austroasiatic language family. Their closest relatives are spoken in Southeast Asia. Vietnamese and Khmer (the main language in Cambodia) are the two largest languages in this family. One possible solution to this puzzle is that the Austroasiatic peoples were the indigenous population of India before the Indo-Aryans swept through north India 4,000 years ago, and even before the Dravidian peoples arrived several thousand years before that. The small pockets of Munda speakers in India’s most remote areas could be the last remnant of that long-lost chapter of India’s history. It is also possible that they arrived in India much more recently.

Like the Muslim-majority areas of Punjab, Kashmir, Bengal, and Sindh, where Hindus were disproportionately present in urban areas, Muslims were overrepresented in Hindu-majority north India’s cities. The United Provinces had seven cities with at least 100,000 people, and all of them had a higher percentage of Muslims than the state as a whole (15%). Its largest city, Lucknow (251,000 people), was 56.7 percent Hindu, 40.5 percent Muslim, and 2.8 percent Other. The next six cities by population were Cawnpore (68.6 percent Hindu, 29.9 percent Muslim, 1.5 percent Other), Agra (61.8, 35.1, and 3.1 %), Benares (68.5, 30.8, and 0.7 %), Allahabad (65.6, 31.2, and 3.2 %), Bareilly (46.8, 52.0, and 1.2 %), and Moradabad (39.1, 57.6, and 3.3 %). Patna, Bihar’s only city with over 100,000 people, was 74.9 percent Hindu, 24 percent Muslim, and 1.1 percent Other.

Below is my map of religion in north India in 1931:

To me, a few things stand out. Obviously, north India was heavily Hindu at the time, but there was considerable regional variation. The northwestern reaches of the United Provinces, what is now Uttarakhand, was nearly 100 percent Hindu. Just to its south, in what is now northwestern Uttar Pradesh, the population was approximately evenly split between Hindus and Muslims and some parts were Muslim majority. The rest of UP and much of modern-day Bihar were solidly Hindu. Several of north India’s large cities stand out as a bit less overwhelmingly Hindu than the surrounding countryside; Agra, Lucknow, and Allahabad in particular caught my eye. The very northeastern tip of Bihar was solidly Muslim. In the south, in what is now Jharkhand, the religious picture shifts from Hindu/Muslim to Hindu/Tribal. Northern Jharkhand blends seamlessly with southern Bihar and was largely Hindu. The two pockets of tribal religion (with a Christian minority) are very noticeable. These areas are also where the Austroasiatic languages I mentioned earlier were spoken. The bright red pocket nestled between the two tribal strongholds is where the Bengali-speaking population of the state was located. This region was transferred to West Bengal after Partition, presumably because it was a more natural fit with that Hindu-majority Bengali-speaking state. When I started making this map, I was most interested in the Muslim population of north India. This was a fairly urbanized Urdu-speaking population, which supposedly made up the bulk of the Mohajirs who migrated to Pakistan in 1947. I wanted to see if there was a noticeable drop in the Muslim population of Uttar Pradesh after Partition. Below is the religious map of north India as of the 2001 Census:

Clearly, Uttar Pradesh did not experience an exodus of its Muslim population in 1947, as the general map didn’t change much from 1931 to 2001. There are a few marginal changes though. First, a few areas in Uttar Pradesh seem to have become more Muslim since 1931, which is presumably a reflection of the much higher Muslim birth rate. Still, the Hindu nationalist hysteria about north India being swamped by high Muslim fertility is not borne out by these maps overall. Second, it appears that some of the cities have become less Muslim. That may be a product of the higher resolution of the 1931 map, but the Mohajirs are generally reported to have originated in the cities of north India, so maybe the urban Muslim population left in 1948, while the rural Muslim population stayed behind (similar to Sindh, where the urban Hindus left and the rural Hindus remained). Third, the northeastern corner of Bihar seems to have become less Muslim (though the borders have changed slightly). There was a population in Bangladesh, called Biharis, who were sort of similar to the Mohajirs. They numbered only a few hundred thousand, and lived mainly in the big cities of the former East Pakistan. After Bangladesh gained its independence in 1971, they were left in a bad situation, because the Bengalis saw them as foreigners, and linked them to the West Pakistanis, and India and Pakistan had no interest in taking them. Perhaps they originated in the northeastern part of Bihar. If so, their absence is faintly noticeable. The fourth point of interest is that about half of the followers of “tribal” religions in 1931 are no longer classified as such. The northeastern cluster of “others” on the first map has vanished. Most of the population seems to be Hindu, though there are a few more Muslims in the area too. I don’t know whether there has been mass conversion of these people since 1931, or if they still have the same religious practices but are not categorized as Hindu. Because Hinduism is a flexible religion, they might have started identifying as Hindu without significantly changing their beliefs.

These are details though. The larger picture is that the religious map of north India has remained largely unchanged since 1931, with the small number of Muslims leaving in 1948 cancelled out by the high birth rate of the majority who remained. India has about five times as many Urdu-speakers as Pakistan, which would imply that about 85 percent of north India’s Muslims stayed in India. So it seems that Partition didn’t affect north India very much, even though the minority of the Muslim population that left played a large role in Pakistan’s early history. Bihar and the United Provinces were both hit by the communal riots of 1946 and 1947, and they linked Punjab in the west to Bengal in the east. So they were certainly part of the larger Partition story, but were peripheral compared to the epicenters in Punjab and Bengal.

Finally, as mentioned above, in my last post, I wondered why Sindh wasn’t partitioned by the British, and I received a number of very interesting responses from readers. Debraj Chakrabarti pointed out that perhaps the better question was why Congress didn’t demand for Sindh to be partitioned. Vikram theorized that Congress might have avoided a partition in Sindh to prevent the Muslim League from demanding Muslim-majority areas in Bihar and Uttar Pradesh. I think they were on the right track. The missing piece, I believe, was why Congress would care whether India had a piece of Sindh versus a piece of Bihar. The answer, I suspect, is in other readers’ comments. Abdul Khair Khan and Indy both made the point that Congress might have wanted the strange lines in the Bengal partition because the Muslim-majority areas of northwestern Bengal were important for Calcutta’s economy. The argument they made is that Congress was willing to swap Hindu and Buddhist-majority areas in less strategic parts of Bengal in order to hold on to the economically important headwaters of the Hoogly River. Combined, these theories lead me to conclude that Congress chose not to request a partition in Sindh because they were afraid that the Muslim League would respond by claiming Muslim areas in Bihar, as well as adjacent Muslim-majority parts of West Bengal containing the headwaters of the Hoogly River. This would have necessitated new partition lines in Bengal that were less favorable to India, although more logical from a purely demographic perspective. Additionally, had Pakistan received northeast Bihar, eastern India would have been physically divided from the rest of the country. I think that explanation makes sense, so thank you to the commenters who helped me gain a better understanding of why the final lines were drawn the way they were.

Why Wasn’t Sindh Partitioned in 1947?

In what is probably the final installment of my examination of the Partition of British India in 1947, I will look at the province of Sindh. I have written previously about Punjab, Bengal, and Kashmir. Bengal and Punjab were officially partitioned along religious lines. Kashmir was subject to a de facto partition, which did not follow the religious divisions of the state. Sindh was not partitioned in 1947, but I would argue that it should be considered in any examination of which country got the more favorable deal in Partition. Sindh also provides an interesting clue as to what might have happened in Kashmir or Bengal had they been left undivided and under Pakistani control. Sindh had a Muslim majority, but several of its eastern sub-districts had a Hindu majority. These sub-districts were contiguous with India, but Sindh was spared a partition.

First though, let’s take a look at Sindh’s demographics as of the 1931 Census. Sindh had about 4.1 million people. About 73 percent were Muslims, 26 percent were Hindus, and 1 percent belonged to other religions, mainly Christianity and Sikhism. In Sindh, the Hindu minority was concentrated in urban areas, while Muslims dominated the countryside. In four of Sindh’s five largest cities, Hindus were an absolute majority. Larkana and Shikarpur, with populations of 25,000 and 62,000 inhabitants respectively, were the two largest cities in northwest Sindh. Larkana was 62.7 percent Hindu, and Shikarpur was 63.5 percent Hindu. Sukkur, in north-central Sindh had about 65,000 people, 58.8 percent of whom were Hindu. Hyderabad, Sindh’s second largest city had 96,000 people and was 70.5 percent Hindu.

The only major Sindhi city without a Hindu majority was Karachi, Sindh’s largest city then and now. Karachi had a very narrow Muslim plurality, 47.8 percent Muslim to 46.6 percent Hindu, with about 5.6 percent of its population belonging to other religions. About 3 percent of Karachi’s population was Christian, and Sikhs and Parsis (Zoroastrians) each constituted 1 percent. This means that Karachi had a non-Muslim majority, which in Punjab determined that an area went with India.

The British record keeping was a bit spotty on divisions within Hinduism and Islam (as opposed to Christianity, which the British documented extensively despite Christians making up only a couple percent of the population). In Sindh, they did record the Sunni/Shia split, and found that 96.1 percent of Sindh’s Muslims were Sunnis. That is surprisingly low, and I wonder if they were able to accurately distinguish between the two sects. After all, it took the British about one hundred years to accurately distinguish between Hindus and Muslims in Bengal.

The final piece of the demographic puzzle for Sindh is language. Sindh’s language landscape in 1931 was different in its nature than the other provinces I have discussed so far. In Bengal, language was not relevant, as Bengali or a closely related dialect was spoken by virtually the entire population. In both Punjab and Kashmir, there were several different linguistic zones. For example, Punjabi was spoken in much of Punjab, but Hindi was spoken in the southeast. Sindh doesn’t look like either of these models. Sindhi-speakers were the plurality in every district, but almost every district had linguistic minorities. Overall, Sindh was 69.9 percent Sindhi-speaking. The two largest linguistic minorities were the Balochi-speakers (7.1 percent), Saraiki-speakers (6.2 percent), and Rajasthani-speakers (3.2 percent). Brahui, Hindi, Punjabi, and Gujarati were all around 2 percent. So too were Kutchi, which is considered a dialect of Sindhi, and Dhatki, a dialect of the very poorly defined Rajasthani language. The Princely State of Khairpur was the only part of Sindh where Sindhi-speakers constituted more than 90 percent of the population. Everywhere else, significant linguistic minorities existed. For example, 29 percent of the Sindh Frontier district’s population was Balochi-speaking, 12.8 percent of Nawabshah district spoke Saraiki, and 6.4 percent of Karachi district spoke Hindi/Urdu. The linguistic and religious demographics of the province have both changed since 1947, but before getting into that, let’s take a look below at the map of religion in Sindh as of 1931:

Unlike Kashmir, Bengal, and Punjab, Sindh had a relatively clean Hindu/Muslim split, with no third group to complicate things. Unsurprisingly, given the fact that Muslims were three quarters of the population, Muslims predominated throughout most of the province. There are three areas in the state where that was not true. Karachi sub-district, in the southwest, was 52.5 percent Muslim (note that I was discussing the city of Karachi above, and not the larger sub-district). Non-Muslims were almost at parity there. The central Hyderabad sub-district was almost perfectly split, with Muslims consisting of 49.4 percent and non-Muslims 50.6 percent. The most interesting area however is the southeast of the province. Four sub-districts, Umarkot, Nagar Parkar, Mithi, and Chachro all had Hindu majorities, and several nearby sub-districts were 40 or 45 percent Hindu. The four Hindu sub-districts combined were about 57 percent Hindu. Below is the same map but with a border added where a theoretical partition along religion lines would have occurred.

I would be very interested to know why these sub-districts did not end up in India. In Bengal for example, one Hindu district ended up in Pakistan, but India was compensated with Muslim-majority districts in a different part of Bengal. In fact, one of the commenters on my Bengal post says that that swap may have been made at the request of the Indian leadership as a way to minimize the economic damage Partition would cause to Calcutta. In Sindh, this was not the case. The only arguments I can think of for not partitioning Sindh are that doing so would have created an undefendable border, that a partition was not worth it given that only a small portion of the province would have changed hands, or that Hindu-Muslim relations in Sindh were so good that a partition was not needed. The first of these is obviously lacking in merit as the British do not appear to have considered this factor in partitioning other provinces. For example, the Punjab line became one of the least defendable borders in the world. It is a line through flat farmland following no geographical logic, and it proved very problematic for Pakistan when India invaded in 1965. So if geography were a consideration, it would have precluded the division of Punjab. Also, much of Sindh’s border with India is a proverbial line in the sand (through the Thar desert). Furthermore, in the far south, the marshy area on the Sindh-Gujarat border has in fact been disputed since 1947, sometimes violently, so the borders drawn in 1947 weren’t clear and logical anyway. The second argument, that giving only the southeast subdistricts with Hindu majorities to India was pointless, is undermined by the fact that most of the Muslim majority district Sylhet, which was part of Assam province, went to Pakistan in 1947, despite the fact that Hindus in the rest of Assam were about as dominant as Muslims in Sindh. The final possibility, that Sindh didn’t need to be partitioned because Hindus and Muslims got along so well, may hold merit, especially if, like me, you are generally against using religion to define a country. However, trusting in good inter-communal relations seems inconsistent with the philosophy of partition, which did not take the Hindu-Muslim relations into consideration, but rather focused on creating separate Hindu and Muslim nations.

My theory is that leaving Sindh united was a de facto compensation for Pakistan, as the lines drawn in Punjab were quite favorable to India, with several Muslim majority regions ending up in India. Now let’s take a look at how the religious makeup of Sindh has changed since 1947. The map below is based on 1998 census results:

Perhaps unsurprisingly, many more Hindus stayed in Sindh than stayed in Punjab. Still, a large number of Hindus left Sindh for India, and their descendants are believed to number three to four million people. There are also about three million Hindus left in Sindh. However, The extent of Hindu flight was not uniform throughout Sindh. As I pointed out above, Sindh’s Hindu population was heavily concentrated in urban areas, and they were the majority in most of Sindh’s cities. According to the 1998 Pakistan census, Hindus now make up only 3.2 percent of Sindh’s urban population, but are 11.6 percent of the province’s rural population. This means that the vast majority of the urban Hindu population left in 1947, but many more of the rural Hindus stayed, perhaps because they lacked to means to leave. An extreme example of urban Hindu flight is Karachi. In 1931, 46.6 percent of Karachi’s population was Hindu. In 1998, Karachi was only 0.9 percent Hindu.

The flight of Sindh’s urban Hindu elite can be linked with the most important effect of Partition on Sindh – that is the arrival of the Muhajirs from India. The Muhajirs were part of the Muslim urban elite of north India who left for Pakistan in 1947. They are still a major force in Sindh, and they are incidentally the only Pakistanis who speak Urdu, which is indigenous to north India, as a first language. They are quite prominent in the arts and business. Former military dictator Pervez Musharraf is also a Muhajir (born in Delhi), though Muhajirs aren’t particularly well represented in the army. For the most part, the Muhajirs didn’t settle in Punjab, which was being ripped apart by religious violence caused by Partition. Instead they went to the very same Sindhi cities that the Hindu Sindhis were in the process of leaving. Karachi, which, as noted, was essentially emptied of its Hindu population, was also the largest destination for the Muhajirs. As a result, Sindh essentially traded one urban elite for another. The impoverished rural Hindu population, however, largely stayed behind. To give a sense of where the Muhajirs live now, here is a map of language in Sindh. Muhajirs, who are Urdu-speakers, are predominant in Hyderabad and Karachi, though they are present in most of Sindh’s major cities. .

The swapping of indigenous Hindu Sindhis for Urdu-speaking north Indian Muslims has had serious implications for Sindh and Pakistan. First, on a macro level, Sindhi culture was impoverished when the urban Hindus left, as any culture that becomes less diverse is. The political weight of the rural Hindus who remained in Pakistan was also reduced, not just because their numbers were reduced, but also because middle-class urban Hindus who could have formed the core of a Hindu voting block in Pakistani politics left for India. More importantly, the Muhajirs who replaced the Hindus played a huge role in shaping Pakistan’s early development. They were closely linked to Muhammad Ali Jinnah (who himself was technically a Muhajir) and his Muslim League. This meant that they dominated Pakistan’s politics from independence until the first military coup in 1958. Their stint as Pakistan’s ruling elite was disastrous. They failed to build democratic institutions, were unable to forge a unified national identity between Pakistan’s eastern and western wings (now Bangladesh and Pakistan respectively), couldn’t gain popular support, and ultimately were swept aside in a coup. The probable reason the Muhajirs failed as a ruling class was that they were essentially strangers in a foreign land. They spoke a different language and had no understanding of Sindhi, Pashtun, Baloch, Bengali, or Punjabi culture. By the time democracy reemerged in Pakistan after the 1971 War, the Muhajirs had withdrawn politically to Karachi, where they still dominate today, leaving the governing of the rest of the country to indigenous politicians from Punjab and Sindh.

Economically the departure of Hindus, both from Sindh and Punjab, was debilitating. In 1947, Hindus dominated commerce, industry, and especially banking throughout most of what became Pakistan. The only profession in which Muslims predominated was leather tanning, which is an industry Hindus abhor for religious reasons. When the Hindus fled, they left Pakistan without its urban middle class. Muslims tended to be farmers, and the Muslim elites were mostly landowners. Eventually, Muslims filled the roles abandoned by the Hindus, but Pakistan would have benefited if the Hindus had stayed and the economic transition had been smoother.

In the aftermath of 1947, Sindh experienced huge ethnic and religious upheaval, which shaped the future of the province in many ways. Karachi, which had been a Sindhi city but was split about equally between Hindus and Muslims, turned into an Urdu-speaking city (though the Pashtun population is rapidly growing), but one that is almost completely Muslim. Sindhi-speakers made up only about 7 percent of the city’s population in 1998, and have probably declined since then. Hyderabad, which was the historic capital of Sindh, is now Urdu-speaking. The countryside remains almost exclusively Sindhi speaking, and has retained more of its Hindu presence. While the demographic upheaval was less dramatic than in Punjab, it was arguably more drastic than was happened in Bengal after 1947. The changes in Sindh demonstrate that it wasn’t only the specific act of partitioning Punjab and Bengal that caused demographic shifts. The fact that the population exchange along religious lines was not contained in Punjab and Bengal, but spread to Sindh, the North West Frontier Province, Kashmir, Uttar Pradesh, Bihar, and Delhi (these last three were where the Muhajirs predominantly originated) argues against an interpretation of Partition as a tragic event of limited geographic significance. Instead, the division of India destabilized the demographic, cultural, and political landscape even in areas that were not directly partitioned. In short, Partition was a tectonic event, which changed the trajectory and the demographic and social makeup of South Asia in a manner similar to World War II in Europe or the Civil War in the United States.

The Demographics of Sri Lanka

So far on this blog, I have focused on the effect of partitions on Punjab, Bengal, and Kashmir. In this post, I’m going to look at a deeply divided country that was never partitioned. Sri Lanka, unlike Punjab and Bengal, is divided along a number of fault-lines. In the cases of Punjab and Bengal, religion was the only factor. In Sri Lanka, religion, ethnicity, and language each play a role in creating the fractured demographic picture of the island. In fact, I can’t think of another society in which all three of these variables are in play. Even famously diverse Lebanon, which I jokingly referenced in the title of this post, is only divided by religion. Using data from the 2012 Census, I will analyze all three of Sri Lanka’s fault-lines in this post, especially in the context of the Sri Lankan Civil War, which ran from 1983 to 2009. Below are maps of Sri Lanka by ethnicity, language, and religion.

Broadly, these three maps are similar, but, as is often the case, the devil is in the details. Note, for example, how the Hindus on the religious map are divided in two on the ethnic map. The Tamil-speakers belong to three different ethnic groups. Sinhala-speakers are mostly Buddhist, but look closer to see the Christian minority on the east coast that emerges on the religion map. I will go into these characteristics of Sri Lanka’s demographics, and more, in the next section of the post. If you aren’t interested in the nitty-gritty details of Sri Lankan religious, linguistic, and ethnic groups, or you already know the basics, feel free to skip this part, and go straight to my analysis of how these demographic factors played a role in shaping  the Sri Lankan Civil War.

A Detailed Look At Sri Lanka’s Demographics

I. Religion

I’ll start with the least relevant, but in some ways most interesting, factor– religion. As of the 2012 Census, Sri Lanka was 70.2% Buddhist, 12.6% Hindu, 9.7% Muslim, and 7.5% Christian. The Buddhist community in Sri Lanka is one of the oldest in the world, going back more than 2,000 years. Sri Lanka’s Buddhists belong to the Therevada branch of the religion, which is the dominant sect in Southeast Asia. Hinduism on the island goes back approximately as far as Buddhism if not further. Sunni Islam was spread to Sri Lanka in the 8th century by Arab merchants, some of whom settled there and married indigenous women. I suspect that there was also a fair amount of conversion, although the Muslims of Sri Lanka maintain the narrative of their Arab origins. Interestingly, even though it has a fairly large Muslim population, Sri Lanka was one of the few parts of South Asia never to fall under Muslim rule. The most recent religious community to emerge was the Christian one. The Portuguese first landed in Sri Lanka in 1505, and ruled much of the island until the Dutch expelled them in 1658. In that time, much of the west coast was converted to Christianity. As a result, the overwhelming majority of Sri Lanka’s Christians are Roman Catholics. There is a small Protestant minority of a few hundred thousand people converted by the Dutch, who stayed until 1796, and the British, who stayed until 1948. Below is the religious map of Sri Lanka:

The religious minorities are concentrated in the north of the country and the east and west coasts. There is also a Hindu area in the center of the island, which I will discuss in more detail when I get to the ethnic map. The Sri Lankan Civil War was not fought on religious lines, although the Tamils are predominantly Hindu and the Sinhalese are predominately Buddhist. Ethnicity was the main driving force. A few characteristics are worth noting. One is the two clusters of Christians. The first cluster is in the north just to the southwest of the main Hindu zone. The second is farther south about midway down the west coast. The other feature I will return to is how interspersed the Hindus and Muslims on the east coast are.

II. Language

The second map depicts the two major languages of the island. I couldn’t find language statistics, but it is well known which ethnic groups speak Sinhala (the Sinhalese) and which speak Tamil (all the others). There is undoubtedly quite a lot of bilingualism, but the goal was to approximate where each language is the “mother tongue.” About 74% of Sri Lanka speaks Sinhala, an Indo-European language most closely related to Marathi and Divehi (spoken in the nearby Maldives). The remaining 26% speaks Tamil, a Dravidian language most closely related to Malayalam and Kannada. It is worth pointing out that the Tamil-speaking population of Sri Lanka makes up only about 10% of the total world-wide Tamil-speaking population. Over 60 million Tamils live in India, mostly in the state of Tamil Nadu, across the Palk Strait from Sri Lanka. The Sinhala-speakers live almost exclusively on Sri Lanka. Below is the language map for Sri Lanka:

When we look at language, all of the complexity of the religious map dissolves and a more binary picture emerges. The religious map shows a diverse and messy island. The language map reveals a country starkly divided by language. The Buddhists and the Christian belt just north of Colombo merge into the orange Sinhala-speaking majority. The Christian pocket in the north, all of the Hindu areas, and the Muslims unify into a solidly Tamil-speaking north, east coast, and center. Looking at this map, it is easier to understand why Sri Lanka spent a quarter of a century locked in a bitter civil war. It also shouldn’t come as a surprise that much of the tension on the island originated in a 1956 bill making Sinhala the national language. Interestingly though, the Sri Lankan Civil War, which in some ways originated in the dispute over language, was not fought along linguistic lines.

III. Ethnicity

In fact, ethnicity, not religion or language, was the main driver of the war. The Sinhalese make up 74.9% of the population. They correlate exactly with the Sinhala-speaking parts of the map above. They are mostly Buddhist, but there is a significant Christian minority. The Sinhalese arrived in Sri Lanka about two thousand years ago, probably from the Bengal area. The Sri Lankan Tamils are the second largest ethnic group, at 11.2%. The Sri Lankan Tamils are mostly Hindu, but again have a Christian minority. The Sri Lankan Tamils were the ethnic group that largely fueled the Civil War, and the rebel groups, including the Liberation Tigers of Tamil Eelam (LTTE), better known as the Tamil Tigers, drew their members from the Sri Lankan Tamil population.  The ancestors of the Sri Lankan Tamils may have arrived at roughly the same time as the Sinhalese, although some argue that they came much later, in the 12th century. It is clear though that whether or not there was permanent Tamil settlement on the island, Sri Lanka’s ties to the Tamil homeland in South India go back millennia.

You may have noticed that the Sri Lankan Tamils only make up around half of the Tamil-speaking population in Sri Lanka. That is because there are two other Tamil-speaking ethnic groups in Sri Lanka, both of which have been subjected to ethnic cleansing campaigns to strengthen the position of one side or the other in the Sri Lankan Civil War and the years leading up to it. The Muslims in Sri Lanka, who as I mentioned above have a different origins myth than the Sri Lankan Tamils, see themselves as a separate ethnic group, despite being Tamil-speaking and genetically identical to the other Sri Lankan populations. They are called the Moors, because the Portuguese arrived on the island shortly after the defeat of the Arab Moors of southern Spain and Portugal and named them after the only Muslims they knew. The Moors make up about 9.2% of the population, and are one of the few examples of a true ethno-religious group– an ethnic group solely defined by its members’ religious affiliation. Unlike the Tamil-speaking Christians and Hindus of the north, the Moors did not support the demand for a separate state (to be called Tamil Eelam), perhaps because of their separate ethnic identity. As a result, they were branded a fifth column and expelled from northern Sri Lanka. The Tamils later apologized, and some of the Moors have returned, although tens of thousands remain internally displaced.

The much-persecuted Indian Tamils make up the final major ethnic group in Sri Lanka. They are descendents of Tamils from India who were brought to the island as laborers by the British in the 1800s. In 1931, they made up 15.4% of Sri Lanka’s population; today they are 4.2%. The Sinhalese-dominated government, starting immediately after independence in 1948, and accelerating in the 1960s and 70s, tried to send as many Indian Tamils as possible back to India. They were stripped of citizenship and forced to return to India. Since most Indian Tamil families had been in Sri Lanka for over a century at that point, they were “returning” to a country they had never seen, and had no desire to live in. This was a transparent attempt by the Sri Lankan government to reduce the Tamil presence on the island. Eventually, the Indian Tamils who had survived the depopulation campaigns of the 60s and 70s were granted Sri Lankan citizenship, as a reward for their loyalty to the government in the Civil War. Still, the Indian Tamils who were expelled have not been able to return. Assuming that Indian Tamil growth rates were similar to those of the Sri Lankan Tamils, this means that the Sinhalese were able to remove about 2 million Tamils from the island, radically altering the demographic balance.

Below is the ethnic map of the island:

A few notable features are the physical separation of the Sri Lankan Tamils from the Indian Tamils. The Indian Tamils are in the center of the island mainly because the hilly interior is where most of the plantations that produce Sri Lanka’s famous tea are located. The Indian Tamils’s ethnic distribution also gives a hint as to why the Sinhalese were so eager to remove the Indian Tamils. If two-thirds of the Indian Tamils hadn’t been deported, much of Sri Lanka’s interior would have been blue on the map above, giving a potential Tamil Eelam more depth. A Tamil state consisting of the north plus the narrow Sri Lankan Tamil strip of land on the east coast would be difficult to defend militarily (as the LTTE eventually discovered). The Sinhalese leadership may have feared that a Tamil-dominated interior would set up a partition in which the Tamils got a large chunk of central and eastern Sri Lanka. Of course, the Indian Tamils remained loyal, making the ethnic cleansing of central Sri Lanka pointless as well as immoral.

Ethnicity, The Tamil Tigers, And Sri Lanka’s Civil War

The question of why the Tamils failed to get a state when so many other minorities, such as India’s Muslims, the Jews of Mandatory Palestine, and the Christians of Timor, succeeded is worth discussing. The basic answer is that the Sinhalese had overwhelming military superiority and a willingness to brutally suppress the Tamils, and no outside power countered that by taking up the Tamil cause. Especially once India disengaged from Sri Lanka following former Prime Minister Rajiv Gandhi’s assassination by an LTTE suicide bomber in 1991, it was only a matter of time before the Sinhalese would use their  military advantage to snuff out the rebellion, which they did in 2009. India could have been Tamil Eelam’s outside backer, much as the British backed the aspirations of Indian Muslims and Jews in Mandatory Palestine.

Instead the LTTE handled India about as badly as possible. India covertly funded them throughout the 80s, and intervened militarily in 1987 to prevent an imminent Sinhalese victory. India was able to force the government in Colombo to agree to a series of humiliating concessions, which gave the north defacto independence, and set off bloody protests against the government in the Sinhalese parts of the island. Every Tamil group agreed to lay down their arms except the LTTE. This was in many ways the turning point for Tamil Eelam, because the frustrated Indian forces on the island tried to defeat the LTTE militarily to preserve what had been shaping up as a smashing victory for India. The Indian “peacekeeping” force became mired in a war with their former clients, and withdrew from the island in 1990. When the LTTE assassinated Rajiv Gandhi, the Prime Minister who had ordered the intervention in Sri Lanka, in 1991, India washed its hands of the whole business, and essentially turned a blind eye to the vicious Sinhalese assault on the north which eventually ended the Civil War.

The LTTE also threw away an incredible opportunity when the Sinhalese government decided to support them against the Indian forces as a way to get the Indians off the island. The government agreed to a ceasefire with the LTTE in 1989. The LTTE waited until about three months after the Indians left before it broke the ceasefire in spectacular fashion, executing hundreds of policemen who had surrendered to it. The LTTE proceeded to expel 70,000 Muslims from the north of the island. In 1993, the an LTTE suicide bomber assassinated Sri Lanka’s President. A pro-peace party was elected in the south in 1994, leading to a ceasefire in January 1995. This ceasefire lasted four months, before the LTTE bombed two Sri Lankan naval vessels.  The LTTE’s actions in the late 1980s and early 1990s were breathtakingly self-destructive. They isolated themselves from what had been a sympathetic India, which was the only country with the ability and interest to make Tamil Eelam a reality, and which had essentially created an independent Tamil state in 1987. They then wasted several opportunities to make a deal with the weak and exhausted Sinhalese, prolonging the conflict and giving the Sri Lankan government and military time to regroup.

While the military decisions by the Tamil leadership were nothing short of moronic, their failure to form a broad tent of Tamil-speakers was perhaps more decisive in their failure to achieve a state. They were unable to reach the Indian Tamils of central Sri Lanka, depriving the rebels of a population situated deep within Sri Lanka that could have put pressure on the Sinhalese population, something the LTTE consistently failed to do. Similarly, the Moors never supported the LTTE, even though it was essentially secular, not Hindu. The effect of this can be seen on the east coast of the ethnic map. The Moors and Sri Lankan Tamils are interspersed all along the east coast preventing a unified Tamil zone there. If the Tamil leadership had succeeded in uniting the Tamil-speakers (and remember the conflict started in part because of the language problem), it could have commanded the support of 26% of the population instead of 12%. Even then, the exact lines of a partition are difficult to map out, because of the distribution of the Tamil-speaking population. A partition would certainly have cause massive population exchanges.  Consider this though. Sinhalese-speakers are divided by religion, Tamil-speakers by religion and ethnicity. Every ethnic group except the Moors are divided by religion. The Christians and Hindus are both split between two ethnic groups. Perhaps Sri Lanka is too fractured for a successful revolt to break out. Paradoxically, it may be Sri Lanka’s divisions that keep it united.

Was There Fraud in Afghanistan’s Elections?

In my last post on Afghanistan’s Presidential election, I remarked that Abdullah was very likely to win the runoff, and then added in passing that Ghani would “need a miracle or massive vote fraud” to win. Perhaps I was too dismissive about the possibility of fraud, given the blatantly fixed 2009 Presidential election. In any case, the preliminary results have been released, and fraud certainly seems to have taken place on a massive scale. The fraud was blatant but largely confined to one area of the country, which should at least make it easier to combat, if anyone in Afghanistan actually tries. First, I’ll post the map of the preliminary results below. It simply shows the official winner of each province; the second map reflects whether the winner of the district won narrowly or in a landslide.

 

If you look back at my original post  back in May, on the first round of the elections, the second round followed largely similar patterns, although a handful of provinces changed hands. Abdullah added Farah province, while Ghani took Kandahar from third-place finisher Zalmai Rassoul, and Kabul, Wardak, Kunduz, and Sar-e-Pul from Abdullah. The other 28 provinces stayed in the same hands, so it is unbelievable that Ghani managed to more than double his vote total between the first and second rounds (2,084,547 to 4,485,888). This wasn’t a rising tide lifting all boats; it was Ghani running up the score in a few provinces he already dominated in the first round. The seven provinces more or less opposite Pakistan’s Federally Administered Tribal Areas (FATA) are the main suspects. These provinces are, roughly from north to south, Kunar, Nangarhar, Logar, Wardak, Paktia, Khost, and Paktika (see here for a map of Afghanistan’s provinces). Unsurprisingly, given their proximity to Pakistan’s Tribal Areas, these provinces are some of the most war-torn in Afghanistan. TTP leader Mullah Fazlullah is believed to be based in Kunar, and Khost is the Haqqani Network’s home base. These seven provinces make up about 16.8% of Afghanistan’s population, and accounted for about 17% of the vote in the first round. In the first round, a hefty 31.9% of Ghani’s total vote came from these provinces. In the second round, 26.3% of the total vote, and 40.8% of Ghani’s vote came from the suspicious seven; 97.6% of the turnout increase between the first and second rounds came from these seven provinces. I couldn’t find the number of registered voters by province, so I used the percentage of the total population that turned out to vote as a proxy. Since children can’t vote and many eligible voters don’t cast a ballot, the turnout should rarely exceed about 40% of the population. I would estimate that any number above 50% of the total population voting is approaching 100% turnout of eligible voters. Below is the map of Afghan provinces by percentage of total citizens who voted in the first round.

And next is the map showing the same variable from the second round:

While turnout in the East increased in general, three provinces, Paktika, Paktia, and Khost stand way out (I imagine even someone who doesn’t know Afghanistan’s provinces could identify these three on the map above). These provinces form the core area of operation for the Haqqani Network and are among the most violent in Afghanistan. The idea that these areas would have “normal” high turnout is a stretch, but what we see here is preposterous. Paktia province had about 64% of its population vote. To give a bit of context, in 2012, about 54% of high-turnout Minnesota’s population voted in the American Presidential election (that would be 76% of eligible voters). Khost province saw turnout reach 73.2% of the population, and Paktika clocked in at a positively inspiring 97.8%. Since over half of Afghanistan’s population is under 18, I doubt that more than 60% of any province is eligible to vote, and that is generous. That means that turnout of eligible voters in Paktika approached 200%, and the other two were almost certainly over 100%. These three provinces netted Ghani about 970,000 votes, or 94.4% of his winning margin.

What I find curious about the fraud is how blatant it is. Even Kim Jong-un only managed 99.97% turnout in his recent “election” victory in North Korea. 200% turnout is hilariously blatant, and was therefore almost certain to be uncovered. Why didn’t Ghani’s people just stuff ballot boxes throughout the country to give him a 10% boost? Abdullah seems to be alleging that Ghani did that as well as the blatant fraud in the east, seeing as he wants 2 million ballots thrown out. It is hard for me to evaluate that claim based on the data I have, though Ghani did do better in the second round across the board. Either there was wider fraud or Ghani experienced a surge in support. Whatever the case is with the potential for wider fraud, I think it is indisputable that massive ballot stuffing in Paktia, Paktika, and Khost occurred in Ghani’s favor. The fact that the massive fraud was concentrated there is interesting. Ghani was already very successful in these provinces in the first round, so the fraud didn’t overturn the will of the voters there; it amplified it. This may explain why there was so little fuss kicked up by local officials. Abdullah cried foul of course, but I haven’t seen anyone from these provinces comment on the fraud. They want Ghani to win. I wouldn’t even rule out the possibility that local leaders stuffed the ballot boxes in an attempt to curry favor with Ghani.

The real question though is whether Abdullah actually won the election or if Ghani won anyway and was just gilding the lily. I don’t know the answer, but I suspect the actual election was very close. If we use the percentages from the second round but the turnout from the first, Ghani wins 51.8% to 48.2%. If we use the second round results but replace the results for seven provinces mentioned above with the first round results, Abdullah would win very narrowly. The other question is whether someone who engaged in blatant fraud on a massive scale should be allowed to ascend to power, even if he won the election legitimately. The fraud shows that Ghani has little respect for democracy and is willing to do anything to gain the Presidency. People like that should be kept far away from power. Imagine if President Obama, having legitimately won the 2012 Presidential election decided, just in case, to fake a few million votes in Philadelphia, Cleveland, and Denver. It is very hard to imagine anyone in a functioning democracy surviving that politically.

A few other thoughts in closing: Last time I wrote about this topic, I discussed how the Persian-speaking part of Afghanistan has the demographic power to dominate the Pashto-speaking parts, and that an Abdullah victory has the potential to tear Afghanistan apart if it is built entirely on Persian voters. The reverse is true for Ghani. If he builds a coalition of Pashtuns and Uzbeks, and adds a million or so fraudulent votes, he could cut the Tajiks and the other Persian speakers out of power permanently. The ethnically-based voting patterns we have seen in Afghanistan are characteristic of dysfunctional democracies. Many voters in third-world countries treat democracy as a way to capture more power for their ethnic or religious group. This has been the case in Iraq and Afghanistan, as well as Pakistan. Last time, I wrote that the Pashtuns would not take kindly to being disempowered, and that they might not accept the results. The fraud in the heavily Pashtun east essentially reflects the Pashtuns not accepting the first round results. Now, the Tajiks are the ones who have been marginalized, and the exact scenario I laid out last time of increased ethnic tension and a north-south civil war could play out, but with the roles reversed.

Anyone who looks at Afghanistan today has to wonder whether this mess was worth a nearly fifteen-year occupation for the United States. All that blood and treasure was spent for what exactly? Ghani, if he ends up becoming President, will have won in part by stuffing ballot boxes in areas of Afghanistan dominated by the Haqqani Network, one of America’s bitterest enemies in the region. If the Haqqani Network didn’t like Ghani, they wouldn’t have allowed their home turf to become a vote factory for him. How can the U.S. maintain a relationship with someone like that? It is a cruel coincidence that, at almost the exact same time that Iraq descends into chaos, and the American adventure there is revealed to be a total failure, Afghanistan too has fallen apart. In this case, the collapse came even before American troops left, which should silence the people who wanted to keep American troops in Iraq (but it probably won’t). In Iraq, the failure of the democracy came because the two main religious groups weren’t able to play nice. In Afghanistan, it has been an ethnic divide. The really worrisome fact for the U.S. is that the failure of Iraq’s democracy presaged the collapse of its security apparatus. In Afghanistan the same thing is likely. At this point, it is hard to imagine Tajiks accepting Ghani as President, or Pashtuns accepting Abdullah. There is some possibility that the two will hash out some kind of compromise power-sharing agreement, but even if they do, Afghanistan’s democracy is dead. Let’s hope the Afghan state isn’t soon to follow.

Updated Afghan Election Results Map

Today, Afghanistan will conduct the runoff for the Presidential elections. I wrote about the first round last month, and I want to add a couple of quick notes before today’s voting. First, since my last post, the final results of the elections have been posted. For the last map, I was working off of preliminary results before the votes had been checked for fraud. I also found a district-level breakdown here, allowing more detail than before. The resulting map is posted below.

Obviously, this map isn’t very different from the original one I posted, but it does demonstrate that Rassoul was very competitive throughout the rural south. He won a number of districts in Helmand and Nimroz provinces, which didn’t register on the first map because Ghani won the overall vote in both. Also, even though Rassoul won Kandahar province, he did not win the city of Kandahar (Ghani did), instead relying on his strong showing in the rural areas of the province. A couple of pockets of support for Ghani in the north also make an appearance. His main support in the north showed up in the two provinces he won, but there is also the small cluster on the border with Tajikistan as well as one around the Wakhan Corridor, which borders China. The people who live in the second area include speakers of Pamiri languages, such as Wakhi, which I mentioned in my last post on the languages of Pakistan. These languages are related to Pashto, which is the dominant language in the south. Also note that three minor candidates who didn’t win any provinces show up on this more detailed version.

The real question is who is going to win today. To me, it seems hard to construct a path to victory for Ghani. Both Rassoul and Sherzai have endorsed Abdullah, and Sayyaf’s vice-presidential candidates have as well. Add those vote totals up, and Abdullah should receive 65% of the vote. This means that Abdullah only needs to win about a quarter of the votes of the candidates who endorsed him. Basically, Ghani would need a miracle or massive vote fraud (always a real possibility in Afghanistan) in order to win. This has the potential to change the way everyone views politics in Afghanistan. Pashtun dominance of Afghan politics has long been taken for granted. The royal family of Afghanistan was Pashtun; the Taliban is Pashtun. Even the puppet leaders installed by the Soviets in the 1980s were Pashtuns. One of the reasons the U.S. decided to back Hamid Karzai is that he was one of the only Pashtuns with anti-Taliban credibility, and American policymakers felt that the new leader of Afghanistan had to be both untainted by Taliban connections and a Pashtun.

However, Pashtuns constitute only about 40% of the population. This means that, in a democracy, they can’t dominate the country as they used to. Karzai, as a southern Pashtun with northern ties, was able to bridge some of the divides in Afghanistan’s demography, but the bottom line is that Tajik rule is probably going to become entrenched. Tajiks, of course, only consist of about 30 percent of the population. This statistic is often repeated by the media, and even used to argue for more Pashtun control. The “Tajik” ethnic designation is very hazy and confusing though. Varieties of the Persian language are spoken in Iran, Afghanistan, and Tajikistan. In Tajikistan, the local dialect is called “Tajiki”. In Afghanistan, it’s “Dari”. In Iran, it’s “Farsi”. For some reason, the speakers of the Dari and Tajik varieties of Persian get lumped into one ethnic group, and the Farsi-speakers into another. I think this likely reflects the Shia-Sunni split between Iran and Afghanistan/Tajikistan. The same is true in Afghanistan, where the Persian-speaking but Shia Hazara are considered ethnically distinct from the Sunni Tajiks. To get a better idea of the number of “Persian” people in Afghanistan and ignore arbitrary “ethnic” lines, the best bet is to look at the languages. In fact, about half of Afghans speak the Dari dialect of Persian as a first language, making them the majority and Pashtuns the minority.

The implications of this split for Afghanistan’s future are very serious. Pashtuns feel that they have a right to govern Afghanistan, but they probably will never be able to as long as Afghanistan’s democracy stands. Obviously, a democracy with issue-based parties that cut across ethnic lines would be ideal, but that possibility is remote in a country where ethnicity is so important. This means that violence and rebellion could be the Pashtuns’ only hope for regaining power. Of course, a Pashtun militant group already exists: the Taliban. That means that we could see a surge in popularity of the Taliban amongst Pashtuns, and a turn towards more explicitly Pashtun nationalist rhetoric from the Taliban. The other factor is that two-thirds of Pashtuns live in Pakistan, and their homeland there forms the base for the Pakistani Taliban or Tehirk-e-Taliban Pakistan (TTP). The possibility of revived Pashtun nationalism is somewhat scary for Afghanistan, but downright terrifying for Pakistan. Afghanistan was the only country to vote against Pakistan’s inclusion in the United Nations because it claimed Pakistan’s Pashtun territory. The North-West Frontier Province (now Khyber-Pakhtunkhwa) was only included in Pakistan in the first place through some fancy footwork and a boycott of the polls by a major anti-Pakistan party. If the current Islamist insurgency in Pakistan is coupled with revived Pashtun nationalism, Pakistan could soon be facing dissolution. This shows why elections can be dangerous for countries where ethnic or religious identity trumps national identity. In these cases, elections are seen as a way to capture power for a particular ethnic group. An election turns into nothing more than a glorified census. Elections accelerate the breakdown of weak states. They do not consolidate them. Whether that means that the West should allow weak states to split up or support dictators who hold them together is a whole different argument, though I would lean towards the former. Today though, we may see the first act of a renewed Afghan civil war, and eventually a serious challenge to the integrity of two of Asia’s weakest states.

Update: See my post on the preliminary results of the runoff here.

What Languages Do Pakistanis Speak? (With Pakistan Language Map)

One of the many frustrations I have faced when trying to understand South Asia is the near total lack of recent data on which languages are spoken and where. The lack of interest in South Asian languages is stunning, especially given that South Asia is home to some of the most spoken languages in the world. The language everyone has heard of is Hindi/Urdu (essentially one language with two scripts), which is spoken by over 300 million people, even if the closely related Rajasthani and Bihari languages are excluded. In the West though, awareness of the other South Asian languages is low. Just to give an idea of how large many of these languages are, here are some comparisons: as many people speak Punjabi as Japanese; roughly as many people speak Bengali as German, French, and Italian combined; as many people speak Oriya as Ukrainian; Pashto has as many speakers as Polish; Marathi, Telugu, and Tamil each has more than three times as many native speakers as Dutch. I have searched for up-to-date statistics for language in India, but haven’t been able to find anything more recent than the 1931 Census. I was able however, to track down tehsil-level date for Pakistan from the 1998 Census. In Pakistan, tehsils are the third level of administrative divisions, after provinces and districts. The data set I found isn’t perfect (more on that later), but it has most of what I was looking for. The data can be downloaded here, and the site also has a link to a pretty cool interactive map.

Before I post the map, I’m going to give a quick rundown on language in Pakistan. English and Urdu are the national languages, and are widely understood, at least by the educated. English, obviously, is nobody’s first language in Pakistan, and Urdu is the first language of about 7% of the population, mostly descendents of immigrants from north India who arrived in 1947. The most widely spoken tongue by far is Punjabi, which is the first language of slightly less than half the population. When Saraiki and Hindko, two Punjabi dialects that are sometimes classified as separate languages, are included, well over half of Pakistanis speak Punjabi or a closely related language. As anyone who read my post on the partition of Punjab will know, a large population of Punjabis (about 35 million) live across the border in India. The second most widely spoken language is Pashto, which unlike Punjabi, an Indo-Aryan language related to Hindi, is an Iranian language. This makes it a relative of Farsi and Kurdish, although Pashto’s closest relatives are a cluster of minor languages known as the Pamir languages which are spoken on the mountainous border between eastern Tajikistan and northeastern Afghanistan. Pashto, like Punjabi, is split between two countries. It is the dominant language in southern Afghanistan, but the majority of Pashtuns live in Pakistan. About 15% of Pakistanis speak Pashto as a first language. Right behind Pashto at 14% is Sindhi, which is a relative of Punjabi. There are a few million Sindhi speakers in India as well, some right on the opposite side of the border, and some Hindus who fled Sindh after Partition. The other major regional language is Balochi, spoken by by about 4%. Balochi, like Pashto, is an Iranian language, though it is not particularly closely related. It is actually closer to Kurdish, leading to the theory that the Baloch may have migrated to their current location fairly recently from the Middle East. Balochi is also spoken in southern Afghanistan and eastern Iran. There are some other minor languages, which I’ll discuss later, but those are the major languages. Below is the Pakistan language map.

Note that I colored the Saraiki and Hindko speaking areas shades of blue because it remains undetermined whether they are separate languages or dialects of Punjabi. Since I don’t speak any of these languages, I can’t make a determination for myself, so I split the difference by making them different shades of the same dark blue. I should also mention two other problem areas. One is the central Balochistan area, which is traditionally considered the Brahui zone. Brahui is a fascinating language. It is Dravidian, which means that it is related to the major South Indian languages, such as Tamil and Telugu, but it is spoken far away from the other Dravidian languages. Brahui barely registers in these data. There are several possibilities. One is that Brahui has lost ground to Baloch. Another is that the Brahui learn both Balochi and Brahui and are equally comfortable in both, leading most to identify the dominant Balochi language as their native tongue. According to some sources, the Brahui have a complicated system of code-switching in which people use Brahui in some situations and Balochi in other situations. Apparently, even within families, there are some times Balochi is used (elder son addressing father), and other times Brahui is used (younger son addressing father). The father speaks to the children in the language of the mother, and wives address their husbands in Balochi. This all seems crazy, but if true could explain why many Brahui would feel comfortable calling Balochi their native language. In any case, it seems that almost all Brahui are fluent in Balochi. Just as a side note, Ethnologue (and Wikipedia) say Brahui is spoken by four million people. This is a ludicrous number, implying that Balochistan, which has 7 million people, is majority Brahui-speaking.

The other problem area was the far north, including northern Khyber-Pakhtunkhwa, Gilgit-Baltistan, and Azad Kashmir. I couldn’t find Census data on Gilgit-Baltistan and Azad Kashmir (which combined make up Pakistan’s part of Jammu and Kashmir). As a result I had to look around the internet for information on these areas. In Azad Kashmir, I had to distinguish between Hindko and Pothohari, another Punjabi dialect being pushed as a separate language (included with Punjabi in this map). It’s a bit difficult to figure out where one begins and the other ends, but it seems that Hindko is spoken in Muzaffarabad, and south of that it is Pothohari. In Gilgit-Baltistan, I was able to use this survey from the early 1990s, which goes into some detail about the northern languages. The other problem is that the languages of northern K-P (Hindko, Khowar, and Kohistani) are all grouped under “other” in my data set. Luckily the geographic ranges of these languages are fairly well known and distinct, so it was easy to figure out which “other”-speaking areas belonged to which language.

I have already mentioned Hindko, but I’ll quickly go through the other six languages that show up on the map in the north. Three of the languages, the aforementioned Khowar and Kohistani, as well as Shina, are Dardic languages, the most northwestern branch of the Indo-Aryan language family. The Dardic languages form an arc in the far north of South Asia. To the southeast in Indian Kashmir, Kashmiri is the most spoken Dardic language. On the other side, in Afghanistan, Pashayi is spoken by perhaps half a million people south of Nuristan province. The other languages in the north are Burushaski (in brown), a language isolate with no known relatives, Wakhi (light purple) which is related to Pashto, and Balti (orange), which is related to Tibetan, and is spoken in Indian Kashmir, though the dialect there is called Ladakhi. The Baltis are almost exclusively Shia; the Ladakhis are split between Shia and Buddhist.

Hopefully this map underscores how linguistically diverse Pakistan is, and possibly explains why the country is so fragmented. Two other features worth noting are the huge swath of northern Balochistan that is Pashto speaking. The 1998 statistics put Pashto speakers at around 30% of Balochistan’s population, but with high birth rates and a surge of refugees from Afghanistan in the last decade, the Pashtun and Baloch populations in the province may be approaching parity. It is also worth noting the tiny presence of Urdu, the national language. While most educated people in Pakistan can speak Urdu, and almost everyone has at least a rudimentary knowledge of it, very few people speak it as a first language. Only the Sindhi cities of Hyderabad and Karachi are majority Urdu speaking. Hyderabad and Karachi were among the only significant Hindu-majority areas of British India that went with Pakistan, and it is possible that the Urdu speakers leaving India went there simply due to the availability of real estate once the Hindus left. Punjab would have been a more logical destination given Lahore’s traditional position as the most important city in northwest India, but Punjab was already overrun with Muslim refugees from India. Sindh wasn’t partitioned, which means it had to absorb fewer refugees. That might explain why the powerful Urdu-speaking community chose the cities of this arid backwater province as their new home.

This map also highlights two large movements for new provinces. The southern Saraiki-speaking Punjab has long had advocates for severing it from the north and creating a separate province centered on Multan. It is unclear how popular this demand is with the average citizen, but the movement has been active since the 1960s and shows no sign of going away. The other potential province would be in the non-Pashto speaking north of K-P. This province would be called Hazara and would be majority Hindkowan (the ethnic group that speaks Hindko).

The final interesting aspect of Pakistan’s linguistic mix is that the border between the Indo-Aryan languages of north India and the Iranian languages runs right through it. This fact, plus the detailed data set I found, gives us the unusual opportunity to investigate the boundary between two major language families. The Indo-Iranian languages form the largest branch of the Indo-European language family. It is typically split into the Iranian branch (Pashto, Farsi, Kurdish and others) and the Indo-Aryan branch (Hindi, Punjabi, Bengali, Marathi and many others). The Iranian and Indo-Aryan languages diverged about 4000 years ago. While South Asia and Iran share many cultural similarities, they are markedly different civilizations. Most of the Iranian peoples share a basic history and culture as do most Indo-Aryans. Below is the map of the border between the Iranian languages and the Indo-Aryan ones.

To me, there are two notable features of this map. The first is the intrusion of Indo-Aryans into central Balochistan. These people are a mix of Sindhi, Saraiki, and Punjabi, which explains why they didn’t register on the first map, since Balochi speakers remain the plurality. Added up though, several tehsils have an Indo-Aryan majority. That corridor between northern Sindh and Quetta is pretty important, because it connects Quetta, and ultimately  Kandahar, to the Pakistani heartland. It is also a major gas producing area for Pakistan. I wonder if the non-Baloch people there are workers who are employed in the gas fields and related industries. That area is also a hotspot for militancy. Perhaps Baloch militants strike there to get at the “foreign occupiers” who are stealing Balochistan’s resources (a common complaint of Balochistan’s active separatist movement).

The second, more macro, feature is the sharp line between the Indo-Aryan languages and the Iranian ones. There are very few parts of Pakistan with mixed communities. This is not at all what I expected. Given that all of these languages, except Urdu, are poorly standardized, I expected the distinctions between them to be hazy. Instead, we see many instances where a 95% Pashto district borders a 95% Punjabi district. This is fairly similar to Western Europe, where the language boundaries tend to be sharp. One doesn’t find many mixed German and Polish towns, or French and Italian. In Europe, most languages are highly standardized and the national boundaries were made to coincide with language borders often through ethnic cleansing. Neither of these is the case in Pakistan. I expected Pakistan’s language map to look a bit more like Southeast Asia’s.

Pakistanis (and Indians) do have very strong ethnic identities. Sindhi speakers know that they are Sindhi and care about the distinction with Balochis. The same is true of Punjabis and Pashtuns. The lack of ethno-linguistic mixing could explain why Pakistan has had such a hard time constructing a national identity. It also could be one of the reasons Pakistan has been so slow to react to the threat of radical Islamic militancy. The vast majority of terrorist attacks in Pakistan happen in Pashtun dominated areas. Since there are few Punjabis or Sindhis living near Pashtuns, those attacks are out of sight and out of mind for the majority of Pakistanis.

Indian Election Results Map

I don’t have that much to add to the conventional wisdom about India’s election results. It was an even bigger landslide for the National Democratic Alliance (NDA) than expected, with the NDA’s leading party, the Bharatiya Janata Party (BJP), winning enough votes to rule without a coalition. This is the first time one party has won enough seats to form a government by itself since Congress after the 1984 elections. It is very likely this will turn out to be a pivotal election in Indian history, and only time will tell whether Congress will recover enough to compete for a majority. The BJP clearly has broken through the ceiling that appeared to cap its share of the vote in the low 20s. I have created a map of the election, which, unlike some of the maps I have seen so far, doesn’t paint all of the NDA parties one color, the UPA parties another, and every other party a third. Every party that won more than one seat gets its own color, which allows one to see that India’s politics remain extremely messy and fractured despite the NDA’s landslide.

First, I notice that Wikipedia has a map up that uses the same base map as me, and even some of the same colors (to be fair, orange for the BJP and blue for Congress is pretty standard). I’m posting this anyway because I have a little more detail with the small parties and because I already made the map. Here are a few of my takeaways:

1. The Regional Parties Fizzled– There was some chatter that a third front would emerge in the space between the unpopular and exhausted Congress and the toxic and divisive BJP, but it totally failed to materialize. Some regional parties did quite well, and overall the non-NDA and non-UPA parties didn’t lose much ground, but the formerly inexorable rise of the regional parties has been halted. For the first time since 1999, the BJP and Congress combined for more than half of the popular vote. By my count, there were six major regional parties that suffered near total wipe outs this year. In two cases, the Communist Party of India (Marxist) losing to the All India Trinamool Congress in West Bengal and the Dravida Munnetra Kazhagam’s drubbing at the hands of the All India Anna Dravida Munnetra Kazhagam in Tamil Nadu, the change is of little import on the national stage.  None of the parties involved were part of either coalition heading into the election. Three parties fell victim to the BJP’s surge in north India. The Janata Dal (United), which won 20 of Bihar’s 40 seats in 2009, won zero seats this year. Similarly, two of Uttar Pradesh’s most important parties, the Samajwadi Party (SP) and the Bahujan Samaj Party (BSP) went from 23 and 21 seats respectively in 2009, to 5 and zero in 2014. Finally, the Jammu and Kashmir National Conference, which has been the dominant party in Jammu and Kashmir since Independence, lost all three of its Lok Sabha seats to the People’s Democratic Party. While some regional parties did fine, the overall trend of the election for smaller parties was negative. In 2009, the BJP and Congress combined for 47.4% of the popular vote and 322 seats out of 543. This year, they won 50.3% of the popular vote and 326 seats.

2. The BJP’s Strength Comes from the Hindi Belt– One  result that really jumps out to me is that much of the BJP’s support comes from the Hindi-speaking parts of India. Hindi is a very poorly defined language, but the broadest definition of the language would include standard Hindi, the various regional dialects in Haryana, Uttar Pradesh, etc., and the languages spoken in Bihar and Rajasthan. If we use this definition, that gives us a Hindi belt containing Rajasthan, Delhi, Uttar Pradesh, Madhya Pradesh, Chhattisgarh, Bihar, Haryana, Himachal Pradesh, Jharkhand, and Uttarakhand. These states hold 225 (41%) of the Lok Sabha seats. The BJP, not including its coalition partners, won 190, or 84.4% of these seats. Add in the BJP’s coalition partners, the number goes up to 201 and 89.3%. The rest of India has 318 seats, of which the BJP won 92 or 28.9%. With coalition partners, the numbers go to 135 and 42.5%. To make the difference more visual, I have two maps below, the first with the non-Hindi areas whited-out, the second with the Hindi belt whited-out:

Hindi Belt

Non-Hindi Belt

Obviously, the BJP did well in the non-Hindi areas, but its dominance was nowhere near as pronounced as it is in the Hindi belt. As to why the BJP did so much better in the Hindi belt, I can only speculate. The north and west of India have always been more receptive to the Hindu nationalist ideology. One possibility is that the Hindi-speaking peoples don’t have an ethnic identity that transcends caste and religion. People who speak Tamil also belong to the Tamil ethnic group. The same is true for Kashmiris, Bengalis, etc. Hindi is one of only a few languages that does not have an ethnic group attached to it. As a result, religious and caste identities play a larger role in the Hindi belt in defining a person’s identity. It is probably not a coincidence that caste-based parties such as the BSP and SP typically do much better in the north, although they did poorly this year.

3. Congress Lacks an Obvious Base from which to Launch a Comeback– The other aspect of this map that I find striking is how comprehensive Congress’s defeat was. Congress didn’t just lose a few critical swing areas, but hold onto its strongholds; it lost nearly everywhere. This wasn’t John Kerry in 2004; it was George McGovern in 1972. Even in areas that the BJP failed to win, Congress lost badly to the key regional parties. Obviously, Narendra Modi could stumble, but it looks like Congress has a lot of work to do even to be competitive on a national level, let alone win a general election. This election showed that all of Congress’s post-Independence goodwill has totally evaporated. If the party wants to win in the future, it will have to move away from relying on its admittedly illustrious history being enough to win it votes by default. I also think that it’s time for the Gandhi family to go. Congress hasn’t had a leader with any real political talent since Indira Gandhi, and even she had a very problematic legacy as Prime Minister. Jawaharlal Nehru was a legend and India’s most important leader other than Mohandas Gandhi, but there is no reason that his mediocre great-grandson should be in control of what is still India’s second largest party. Congress needs to become a true left-wing party to provide a counterweight to the BJP. It should not continue as a vehicle for members of the Gandhi family who have nothing better to do with their time.

Finally, I just wanted to consider how Indian politics might look now if Partition had never occurred. It has been argued, persuasively in my opinion, that Partition created the conditions necessary for radical Islam to flourish in Pakistan. Obviously, it is too soon to tell if Narendra Modi will end up moderating like Atal Vajpayee, the last BJP Prime Minister, or catering to the radical Hindu fringe of his party. Still I wonder if Partition didn’t create the necessary conditions for Hindu fundamentalism as well, but the effects have been delayed by luck and better leadership in India. Hindu-Muslim relations in north India have always been worse than in the rest of the country, and India as currently constituted is politically dominated by the north. I wonder if an undivided India, with 500 million Muslims instead of 180 million would have elected a party like the BJP. Right now, Muslims are too geographically dispersed to be a powerful voting block. But the BJP won only 9% of the Muslim vote, according to this New York Times article. If they’d had to compete in Lahore, Peshawar, or Chittagong, that level of Muslim support probably would have been crippling to Modi’s Prime Ministerial hopes. It has been clear for years that a Pakistan ripped away from the traditions of South Asia and reoriented towards the Muslim world will suffer from a lethal combination of extremism and rudderlessness. I wonder if a democratic India, shorn of two thirds of its Muslims, lacks a necessary check on the authoritarian tendencies of its Hindu nationalists, but we’re only just starting to realize it.

Update 07/30/2014: When I first wrote this post, I neglected to mention how Modi’s election could change India’s relationship with the United States. To me, it isn’t yet clear what effect the new Prime Minister will have. President Obama apparently had a strong personal relationship with Manmohan Singh, but the U.S. disengaged from India under Obama, possibly because it was clear that the Congress government would not recover from numerous corruption scandals and was essentially living on borrowed time. Now that the Indian government has a strong mandate, perhaps the U.S. will seize the opportunity to recommit to the relationship with India despite the fraught history the U.S. has with Narendra Modi. Modi was the subject of a travel ban in the aftermath of the 2002 riots in Gujarat, but that ban has been lifted, and Modi is expected to visit Washington D.C. in September. On the American side, Secretary of State John Kerry will travel to India this week and Secretary of Defense Chuck Hagel will follow next week. Modi has billed himself as an economically focused leader, which could lead to improved economic ties with the United States if, as promised, he passes reforms that make it easier to do business in India. In the short-term, the U.S. is likely to be cautious with Modi, trying to balance the need to strengthen one of the U.S.’s most important relationships in Asia while also remaining wary for now about Modi’s checkered past and questionable commitment to multiculturalism and democratic values.

Elections, Elections! (Afghanistan Election Results Map)

South Asia is currently in the midst of two momentous elections. Afghanistan wrapped up its first round of voting in early April, and the preliminary results were released a couple of days ago (update: official first round results here). India’s elections are in progress, with Punjab, Gujarat, the part of Andhra Pradesh that will become Telangana, plus parts of Uttar Pradesh, Bihar, and Kashmir voting today. Because we don’t yet know the results of India’s elections, I’m mainly focusing on Afghanistan in this post, but I have some thoughts about the Indian election at the end. The first round of the Afghan election failed to produce a winner, as no candidate exceeded 50% of the vote. This means that the top two vote-getters, Ashraf Ghani Ahmadzai and Abdullah Abdullah, will face each other in a runoff in late May or early June. Before we get into any analysis, here are some maps of the election results. This first map shows the winner of each province:

For a more detailed look, the next map shows the colors in proportion to the percentage of the vote each candidate won, to distinguish between a landslide and a narrow win:

Notice for example that while Ghani won the south and the east, he dominated in the east, but eked out most of his wins in the south. To me, this map yields several interesting pieces of information:

1. Karzai’s Clout- President Hamid Karzai was widely seen as supporting Zalmai Rassoul, a southern Pashtun. Karzai’s brother Quayum withdrew from the race and supported Rassoul, making Rassoul the likely proxy for Karzai. Rassoul never really had a chance to make the runoff, but how well he did could indicate how much clout Karzai has left, and how much influence he will retain when he leaves office.

With the results in, things don’t look good for Karzai. Rassoul netted about 11.5% of the total vote, finishing a distant third, behind Abdullah (44.9%) and Ghani (31.5%). He did win Karzai’s home base of Kandahar, but he was shut out everywhere else. Rassoul only surpassed 20% of the vote in a few southern provinces: Urozgan, Zabul, Helmand, and Nimroz. He received 8.4% of the vote in Kabul. This would seem to indicate that outside of Kandahar, Karzai’s support and influence have largely eroded, even in the south. I wonder if these ugly results for Rassoul will change Karzai’s thinking on whether to leave Afghanistan when his term is over or even seize power by declaring a state of emergency.

2. A Tajik President?- If Abdullah wins the second round, he would be one of the very few non-Pashtuns to lead Afghanistan. Pashtuns are the largest ethnic group in Afghanistan, consisting of roughly 40% of the population. Abdullah is half Pashtun and half Tajik, but he is identified mainly as a Tajik and he was a member of the Tajik-led Northern Alliance in the 1990s. Afghanistan has been led by Pashtuns for almost all of its history, with one notable exception being Burhanuddin Rabbani, who was president in the 1990s between the fall of the Soviet puppet government and the arrival of the Taliban. The question about an Abdullah win is whether the Pashtuns would accept a northerner, especially one with close ties to Ahmed Shah Massoud, as President. Ethnic tension in Afghanistan has been relatively low since the American intervention, but if Abdullah fails to convince the Pashtuns of his legitimacy, the stage could be set for a north vs. south civil war to reignite.

3. Where were the Uzbeks?– One of the more mysterious results of the election was Ghani’s relative weakness with Uzbeks, who make up about 9% of the population. His top Vice President (they get two in Afghanistan) is Abdul Rashid Dostum, an Uzbek warlord who fought the Taliban in the 1990s, sometimes as part of the Northern Alliance. He was selected to ensure that Ghani had at least one ethnic group other than the Pashtuns in his corner. Ghani won two northern provinces, Juzjan and Faryab, but failed to win several other provinces– such as Samangan and Kunduz– that are thought to have large Uzbek populations. There are several possible explanations for this. One is that there was vote fraud in those provinces, and Abdullah didn’t really win them. Two is that Dostum only influences the western Uzbeks in Afghanistan. Three is that many Uzbeks didn’t show up to vote. Four is that there aren’t as many Uzbeks as is usually assumed.

To me, the fourth explanation is the most logical. Afghanistan has never done a full Census, with the most serious attempt in 1979 interrupted by a coup, civil war, and the Soviet invasion. The most common map of Afghanistan’s ethnic makeup, seen here, is of unknown providence (to me at least), but it very well could be decades old. This map, which is not the usual map shown for Afghan ethnic groups, seems to dovetail better with the election results, and it shows the Uzbek area as much more limited in size. My other reason for leaning towards a smaller Uzbek population is that there is no explanation that makes sense for why one of the first three options is right. I have seen no indication that Dostum only holds sway of part of the Uzbek population, that voter fraud was particularly bad in the north, or that half of the Uzbeks didn’t vote while the other half did. If indeed the Uzbeks are fewer than previously thought, or live in a more concentrated area, then the ethnic makeup of the north needs to be reevaluated, and the strength of the Tajiks may be greater than everyone has assumed for years.

Who will win?– A lot depends on who the also-ran candidates endorse, but Rassoul or fourth place finisher Pashtun warlord Abdul Rasul Sayyaf could put Abdullah over the top if they swing their votes to him. Unless Ghani manages to rally all of the non-Abdullah candidates to his side, he will have a very tough time making up the lost ground. Even if he does, Abdullah could win on his own, especially if the Uzbeks are less influential than Ghani calculated and turnout is low in the Pashtun south. So I think that Abdullah is the likely winner. The real question is how close the runoff will be and how much trouble the Taliban causes. Neither one of these is knowable right now, but a very close result would open the door to claims of fraud from the loser, and Taliban attacks could de-legitimize the results or even goad Karzai into calling a State of Emergency and holding on to power.

A few thoughts on India’s elections– India’s elections are still ongoing, so it is hard to come to any conclusions yet, but some trends seem to be emerging. One is that turnout is fairly high, about 5-10% higher than in 2009. This is typically reported to favor the BJP, and Narendra Modi and the BJP seem to be doing very well. Polls have shown the BJP-led NDA getting about 100 more seats than Congress’s coalition the UPA, but the most recent polls have indicated that the BJP’s margin of victory could be even greater than that. There is even some possibility that the BJP won’t need to expand its current coalition at all to form a government.

There also appear to be some big shifts coming at the statewide level. The statewide polls show that Congress, which won 33 out of 42 seats in Andhra Pradesh in 2009, will essentially get wiped out in the state (or states, as Andhra is set to be bifurcated soon). The breakaway YSR Congress and the Telangana-based TRS are set to win 5-10 seats, and the BJP’s partner in the state is polling in the high teens, leaving Congress with around 5 seats in what had been one of its most important states. In Bihar, Congress and its coalition partners could actually gain seats, with the Janata Dal (U) looking like the BJP’s main victim. The JD(U) won half of Bihar’s 40 seats in 2009, but it could win as few as two or three this time. The BJP is also poised to nearly sweep two huge states, Gujarat and Madhya Pradesh that it won narrowly in 2009. It also seems possible that Congress will lose all seven of Delhi’s seats, which it swept in 2009. Congress won 20 of Rajasthan’s 25 seats in 2009 and may only hold on to two or three this year.

The biggest shift appears to be coming in the massive northern state of Uttar Pradesh (population about 200 million). In 2009 Congress won 21 of UP’s 80 seats, with the BJP winning 10. The two main regional parties, the Dalit-focused BSP and the Samajwadi Party have typically been very powerful. In 2009, the BSP won 20 seats and the SP won 23. The BJP may be on the verge of establishing itself as the biggest party in the state. Congress is heading towards the low single digits, and the BSP and SP are in the low teens. That means that the BJP could win 40 or even 50 seats in Uttar Pradesh, which would be an incredible achievement in India’s fractured electoral landscape.

Of these results, the worst for Congress, in my opinion, is the wipe out in Andhra Pradesh. Congress has lost the state only twice in its history, and fought hard to create the new state of Telangana as a way to hold on to some support there. Only winning a handful of seats would be very bad. But really, none of the individual states are deadly for Congress, as Indian elections are notoriously swingy, and states often move against the prevailing political winds. The real problem is that all of these debacles are occuring at the same time, with precious few bright spots in other parts of the country. In other words, things look very very bad for Congress and very very good for the BJP.

Update: See my updated posts on Afghanistan’s elections here and India’s elections here.

Update 2: See my post on the preliminary results of the runoff here.

Where Does Sectarian Violence Occur in Pakistan?

Sectarian violence in Pakistan is seen as a second tier problem by many politicians in Pakistan, partly because most sectarian groups pose no threat to the state, and partly because Shia are not particularly well-loved in Pakistan, which has about four times as many Sunni as Shia. For Americans though it is one of the easiest issues to understand, due to years of reading about sectarian violence in Iraq. I mapped sectarian violence by district in Pakistan since 2010 using the database on the South Asia Terrorism Portal (SATP). SATP ignores violence against Christians and only includes one attack against Hindus, instead focusing on minority sects of Islam. So I separated all of SATP’s sectarian attacks since 2010 by district and mapped them.

The results were somewhat underwhelming. Basically sectarian violence is a major problem in Karachi and Quetta, but rarely occurs elsewhere. Karachi alone accounted for about 55% of all sectarian attacks since 2010, with Quetta accounting for 19%. The remaining 26% was spread pretty uniformly over the rest of Pakistan. Below is the map:

The takeaways are that most of Pakistan sees very low levels of sectarian violence, and two cities see very elevated levels. The real question is why Karachi and Quetta are such problem areas while the rest of Pakistan is mostly free of sectarian conflict. The answer in Karachi could be linked to the generally unstable security situation in the city and the spectacular diversity that ensures contact between different communities. Sectarian groups might be attracted to the Karachi’s lawlessness and plentiful targets. My theory for Quetta’s sectarian violence is that the Lashkar-e-Jhangvi (LeJ), a nasty sectarian group with deep roots in Pakistan, has been given free rein in northern Balochistan by the TTP. The TTP isn’t particularly interested in Balochistan, so they outsourced their operations there to the LeJ, who have naturally continued their sectarian ways, but now with TTP support. Quetta also has a large community of Shia Hazara refugees from central Afghanistan who are often targeted. In any case, Karachi and Quetta are exceptions to a rule of generally decent relations between Shia and Sunni in Pakistan. In the end, it looks like sectarian violence in Pakistan is not as widespread or serious as I would have expected.