The ethnic/religious/regional composition of a state’s security forces is incredibly politically important. Political scientists, from Horowitz to Enloe Petersen to Cederman et al. to Bellin to Roessler to, most recently, Johnson and Thurber (among many many others) have highlighted ways in which the composition of the state can affect political stability, coups, counterinsurgency/internal security posture, and revolts. Yet for many obvious reasons, military and security forces tend to be extremely reticent to share this kind of data publicly, so studying it is difficult.
In the South Asian context – one of the inspirations for Horowitz in particular (especially Sri Lanka – see also his less-well-known book on the attempted coup of 1962) – we have recent-ish work touching on the demographic composition of the Pakistan Army by Fair and Nawaz (at the level of recruitment intake) and by Dann Naseemullah, Ahsan Butt, and me (at the level of the corps commander tier). A related, excellent, recent ethnographic book on the Pakistan Army was just published by Maria Rashid. Classic comparative-historical work by many, such as Siddiqa, Cohen, Shah, Jaffrelot, Rizvi, Jalal, and Fair, has highlighted ethnic imbalances at a macro-level within the military apparatus.
On India, Steven Wilkinson has unpacked Army data on regional patterns, and I’ve worked with Drew Stommes to tentatively back out spatial variation in force composition in the BSF and CRPF using published fatality data. Omar Khalidi focuses in more on the police.
The newest contribution comes from David Smith in his Stimson Center monograph The Wellington Experience (2 years ago I highlighted his counterpart monograph on Pakistan, The Quetta Experience). On page 49, Smith offers a summary of some of Wilkinson’s findings for those not familiar with it.
The monograph further provides (highly tentative) data allowing insight into Muslim and Sikh representation in the Indian Army’s higher ranks (p. 48):
Annex H is on page 248 and is crystal-clear to caution that “these figures are not official and should be taken as illustrative only,” so massively important caveats abound. But this is nevertheless a valuable addition to this very challenging empirical agenda.