A Spatial Analysis of War Data in Pakistan – Khyber Pakhtunkhwa and FATA (II – Monthly Data)

An Introduction to this Series on the years 2009 – 2012 in conflict in Pakistan can be read here. This section deals with data available on Khyber Pakhtunkhwa and the FATA on a monthly scale. We looked at the annual scale earlier and now look at monthly data, trying to see where influences between KP and FATA can be judged.

You may als want to read on where we see the limits of such an analysis and to what extent we are qualified in this field.

All Data is from the SATP Portal.

What do we expect from a monthly analysis of this data?

A number of interests are at play in the War in Pakistan, with especially many players in the FATA and the bordering districts in KP.

• Locals have an interest for peace and prosperity and politics in which they have a say – where they do not get that or are supressed, they do at times use violence to defend their ineterests against rival groups or the Pakistani National or Provincial Government and its loosely attached arm, the Army and Security Forces.
• Non-Local Insurgent Groups fight for influence and power, sometimes against local groups, sometimes together with them.
• International Terror Outfits use the FATA as a safe haven and a staging ground for their war with its multiple enemies (the Western Troops, the US, the Pakistani and Afghan State).
• Non-Local Groups from elsewhere in Pakistan do seem to have close links to the FATA at times.
• The US is leading a war against international terror outfits, with little interest seemingly in how that affects ‘their’ wars – the conflicts in Pakistan proper. Their aim is not so much to decrease violence in the (wider) area but to decapitate these outfits.
• The Pakistani State and Army are waging a war against local and international outfits, while it is sometimes unclear who they target and who they let prevail.

The first group is largely a victim in this war. If they surface as a perpetrator of violence, it is in armed clashes (Lashkars, Tribal Militias). The next three are actors and responsible for blasts (suicide attacks, bomb blasts) as well as armed attacks and defence against the Pakistani Army as well as International Forces over the border to Afghanistan. The fifth is in charge of the Drone Strikes and the last is responsible for most armed encounters.

Ideally, a higher temporal resolution of this war data will enable to judge how the actions of these actors do influence and trigger each other and whether they rather escalate actions from the adverse side or disrupt action. Pakistan Army Operations and Drone Strikes should ideally lead to decreasing attacks or at least less intense blasts.

By looking at districts and its neighbours, we should also be able to judge how violence spills over district borders. We are doing this here solely based on the data, not on considering the different insurgent factions and their possible links which need not be geographically adjacent. Once we have judged from the data, will go back and see whether these observations can be backed by known affiliations, locations and tactics of different outfits as well as tactics of the Pakistan Army or the Drone Program.

FATA and KP Frontier Region – Monthly Data

Looking at annual data limits the possibility of judging whether blasts occured as a response to Army interventions or drone strikes. Even on a monthly scale that is often still not possible. However some trends become obvious. We have looked at all FATA individually and included the KP districts of DI Khan and Tank (Southern Waziristan Area), Bannu, Lakki Marwat and Karak (Northern Waziristan Area), Hangu, Kohat and Peshawar (Central Area) and Dir and Charsadda (Northern Area) with the respective FR added to the KP side to see how conflict has spilled over the borders of the FATA or in the other direction. All FATA data is hence without the FR, but these are included with the respective districts in KP.

In Figure 1 (and subsequent for each Agency) blasts, drone strikes and armed encounters are compared. Also blast intensity (killed and injured per blast) is included to have a measure for the scale of the attack.

Looking at all FATA, one can make out a relation especially between two time series – armed encounters, which mostly include attacks of the Pakistan Army against insurgents and Blast Intensities. As soon as armed encounters drop, huge bomb blasts are the result (especially visible in January 2010, September 2010, November 2012, January 2012 and November 2012). At a smaller scale this can also be said for drone strikes and number of blasts. However, since drone strikes do not cover the FATA evenly, this will be looked at in detail for the respective FATA.

For Bajaur (1b) blasts and army interventions do go hand in hand (R=0.44). In no other Agency is the relation between these two that high. However it is not so clear whether the first sparks the latter or vice versa.

In the Khyber Agency (1c) intensities of blasts do seem to rise just after or during military surges. Also blasts do seem to occur once military interventions intensify but they do so with monthly breaks (December 2010 to August 2012). In times where the Army is less active, blasts continued unabated every month (May to December 2010).

In Kurram, relatively few armed encounters seemed to keep blasts rare during 2010 and 2011. Since actions here have largely dropped the number of blasts have risen in 2012 and so has the overall intensity.

Mohmand (1e), has experienced many and violent attacks at the beginning of 2010, when army operations were minor. With the beginning of the Mohmand Offensive (Brekhna) blasts surged at the end of the year but then died down. Since operations have dwindled blasts have picked up again during 2012.

The North Waziristan Agency (1f) has been the most heavily targetted by the Drone War but has seen relatively few attacks. Whether that is because of the drone attacks or the lack of Army Intervention in the area is difficult to tell. Also drone strikes do not seem to call for immediate responses in the geographical vicinity, nor seem Pakistani Army interventions in any way coordinated with drone strikes.There is an indirectly proportional relation between drone strikes and blasts and their intensity (R=-0.36/-0.22), contrary to a directly proportional one to blasts’ intensity in the neighbouring districts (R=0.25). Armed Encounters and Blasts in the Agency are however directly proportional (R=0.25).

Orakzai on the other hand (1g) has seen nearly no drone strikes but enourmous fighting troop presence. Equally, blasts occur rather seldom, but they do seem to occur especially during Army interventions (R=0.28). For both Agencies the spill over effect into neighbouring areas is of importance.

South Waziristan was the theatre for a massive army offensive coupled with numerous drone strikes in 2009 which was followed by violent attacks. Increasing drone strikes in mid-2011 and a number of clashes have resulted in more blasts. Similarly to North Waziristan, the drone strikes have a diminishing effect on blasts in the Agency itself but increase blasts in the neighbouring districts in KP (R=0.33). Armed Encounters and Blasts in the Agency are however again directly proportional (R=0.28). The situation has since 2011 been rather stable, except for the increase of intensities at the end of 2012.

Spill Over Effects – Links to KP

We have now looked at the respective neigbouring districts in KP, to see how actions of war have shown effect in these areas. We have not looked at the relation between the different FATA yet, because relations between the Agencies are likely not driven foremost by geographical proximity but rather by tribal and political alliances. These we need to investigate before we attempt an intra-FATA analysis. The neighbouring districts, irrespective of tribal or political alliances are known to bear the brunt when groups flee army interventions in the FATA themselves. This has happened between Bajaur, Mohmand and Dir (all the way to Swat) respectively and is the case for Orakzai and Hangu/Kohat.

Figures 2a – 2e compare blasts and their intensity in the respective intensities in the neighbouring districts of KP (including the FR of DI Khan, Tank, Bannu, Kohat and Peshawar) with drone strikes and armed encounters in the Agencies.

While until late summer 2010, the number and intensities of blasts have largely been a response to increasing activities of the Pakistani Army and have spiked when the army retreated immediately (Dec 2009 – Jan 2010) or dropped when the army stayed (May 2010), from September 2010 the intensities of blasts have nearly continuously increased with increased drone strikes, and decreased again when drone strikes dropped. Since the drone war is probably little influenced by attacks in Pakistan, it can be assumed, that these attacks happen as a response to drone attacks and not vice versa. While such a direct response is not apparent in FATA itself (see Figure 1a) it looks like the response to drone strikes but also forays of the Pakistan Army really needs to be looked for outside the Tribal Agencies.

Intensities have overall rather dropped, but the number of blasts has stayed largely the same.

Assuming that there is a relation between outfits based in Mohmand and Bajaur Agencies and attacks in Dir and Charsadda (a rough assumption, attackers in these districts may well come from Khyber, Orakzai, from Afghanistan or further South), one can observe how attacks spike once army interventions drop (April 2009, November 2009, February 2010, February 2011, September 2012).

In the Central FATA (Kurram, Orakzai, Khyber) offensives by the Pakistan Army have been especially numerous and so have been attacks in Hangu, Kohat and Peshawar. Blasts here occurred seemingly in 4 month-cycles, somewhat indirectly proportional to army interventions. It’s not clear whether this is solely because it takes so much to organise an attack again or whether that is the time that is suitable to let Security Forces become slacking again after a major attack.

In the neighbouring districts to the North Waziristan Agency, where drone strikes are frequent and ample, the intensity of bomb blasts and suicide attacks is generally rather high and blasts do seem happen more often and with higher intensity after numerous drone strikes (2d).

In the area of South Waziristan events are overall at a small level with trends difficult to observe visually. However it seems that attacks seem to increase intensity around the times of drone strikes in the neighbouring Agency (2e).

Conclusive Remarks

It seems that there are some prevalent traits in the data. Drone Strikes in a FATA agency cause attacks in the same Agency to decerease at the same time, while they significantly increase attacks in neighbouring districts in KP. Army Interventions in all Agencies cause attacks to increase in the same Agency however and only interventions in Bajaur and Mohmand have caused attacks in the neighbouring Dir and Charsadda to decrease at times (R=-0.16). Drones hence have to seem a ‘better’ effect in the FATA itself, but have a stronger spill over effect.

Just from the data, little relation can be found between the Agencies. Army Offensives in Orakzai have perhaps caused attacks in South Wziristan to decrease (R=-0.32), which would be explainable with a close relation between the two Agencies (many insurgents from South Waziristan fled to Orakzai when the Army started their offensives there). Armed Encounters in Kurram and Attacks in Khyber are furthermore similar (R=0.4). No other time series show any resemblance over the 4 year period.

Before any more conclusions can be drawn, we will look at the tribal, political and tactical relations between the regions.