Are constituencies in Bihar cast in stone? | Number Theory

While the caste-survey data is extremely useful in making state-wide analysis, it has an important limitation because no granular data was published in report

These pages have dealt with the importance of caste in Bihar’s political competition in detail on the basis of two databases – a historical database of Bihar’s 3629 MLAs from 1962-2020 and 498 candidates of the two major alliances in the 2025 election – prepared by the first author of this story. They have also married these statistics along with other statistics on Bihar to build a larger political economy narrative in another three-part series. This analysis was facilitated by the fact that Bihar is the only state which has disaggregated data of a recent vintage at the sub-caste or jati level thanks to the caste survey conducted by the Government of Bihar in 2023. While the caste-survey data is extremely useful in making state-wide analysis, it has an important limitation because no granular data, even at the district level, was published in the report.

AFP picture

• How does caste matter in electoral competition at the grassroots level?

  It is a fair assumption to make that the social composition of the population differs significantly across Bihar, as it would in any other large state. Political competition, in an assembly election, is decided at the level of an assembly constituency (AC), of which Bihar has 243. How many of these ACs are dominated by a particular social group or sub-caste? One way to answer this question is to look at the social background of MLAs in a given AC across elections. Replicating this exercise for the entire state can give useful insights into how caste-fixated Bihar’s ACs are. This two-part series will try to execute this analysis by scrolling through HT’s historic database of Bihar’s MLAs from 1962-2020. The first part will lay down the structural caveats to such an analysis and the second part will look at trends in Bihar’s ACs being dominated by a particular caste.
• 
  Strictly speaking, ACs cannot be compared over longer time periods

  HT’s caste database of Bihar’s MLAs covers 15 assembly elections from 1962-2020. However, this does not mean that all 243 ACs which exist today have had 15 iterations too. In India, periodic delimitation exercises redraw AC and parliamentary constituency (PC) boundaries. The last delimitation was undertaken in 2008 and Bihar has had three such exercises (1963, 1973 and 2008) during the period covered in the HT dataset. Not only does delimitation redraw AC boundaries, it also discontinues existing ACs and creates new ones. What it also does is redesignation of ACs and PCs into unreserved or Scheduled Caste (SC) and Scheduled Tribe (ST) reserved. This matters while counting the caste of the elected representative because reserving or de-reserving a constituency can change the caste composition of candidature there. How do all these things play out in HT’s database of MLAs, which is the basis of this analysis? There are 343 unique AC names in this dataset. Only 151 of them consistently appear in every election between 1962 and 2020. To be sure, even these might not be strictly comparable because of some redrawing of AC boundaries. There are 59 ACs which saw a change in their unreserved/SC-ST reserved status.
• 
  After the 2008 delimitation, almost half of Bihar's ACs have had an MLA from the same sub-caste

  Comparing the sub-caste of Bihar’s MLAs across ACs in the 2010, 2015 and 2020 assembly elections is the most water-tight statistical comparison which can be made because AC boundaries have not changed during this period. HT’s analysis shows that 107 out of Bihar’s 243 ACs have returned an MLA from the same sub-caste in these three elections. 87 of these are unreserved ACs while 20 of them are SC/ST reserved. What is even more revealing is the fact that the winning party was unchanged in only 48 of these 107 ACs. This shows that caste loyalty often triumphs over party loyalty in the state. The sub-caste which has the largest share in these ACs is Yadav. However, of the 27 such Yadav stronghold ACs, only 13 ACs have been won by the same party in all three elections The sub-caste which has the largest share in these ACs is Yadav. See Map 1 (same caste in last 3 polls shaded) & Chart 2 While comparing ACs from 2010, 2015 and 2020 elections is the strictest criteria to track ACs which are caste strongholds, there is some merit in doing a long-term comparison as well even though AC boundaries might have changed somewhat. However, this is unlikely to have changed the fundamental nature of the electorate on a caste-composition basis. This is what the second part of this series will look at.