Vote Geography, Wasted Votes, and AIMIM’s Seat Bonus

Under first‑past‑the‑post (FPTP), AIMIM’s spatially concentrated vote turned into seats far more efficiently than TRS or BJP, a pattern you can see by votes‑per‑seat, seat–vote gaps, wasted‑vote accounting, and the distribution of winning margins.

Quick facts

*Efficiency ratio = seat share ÷ vote share. Values >1 indicate over‑representation relative to votes.

Source of the underlying data: Wikipedia

Why conversion differs

Two features of the Hyderabad map matter. First, geographic concentration: AIMIM’s support is clustered in the Old City. In FPTP, clusters spanning many wards create multiple wins with relatively modest vote totals per ward. Second, two‑way competition elsewhere: TRS and BJP contest across much of the city, splitting the field and producing many narrow wins and losses. Diffuse support plus close races raises the number of votes that do not produce seats.

Sorting the ward‑level results by margin, the largest twenty winning margins are all AIMIM—textbook packing. That coexistence of many extremely safe wins and very few votes outside the cluster is precisely what generates the seat bonus.

Wasted‑vote lens

Wasted votes definition by ward w for party p:

• If p loses: All v_p,w are wasted
• If p wins: Wasted votes = v_p,w − (runner-up vote in w + 1)

Formal expression:

$$W_p = \sum_{w \in L_p} v_{p,w} + \sum_{w \in W_p} \left[ v_{p,w} - \left( \max_{q \neq p} v_{q,w} + 1 \right) \right]$$

Where:

• $W_p$ = total wasted votes for party $p$
• $L_p$ = set of wards where party $p$ loses
• $W_p$ = set of wards where party $p$ wins
• $v_{p,w}$ = votes for party $p$ in ward $w$
• $\max_{q \neq p} v_{q,w}$ = runner-up's vote total in ward $w$

AIMIM’s geography—few losses and many lopsided wins inside the Old City—keeps $\frac{W_p}{\sum_w v_{p,w}}$ comparatively low versus rivals that spread votes thinly and lose many close races. This reconciles the top‑20 margins all AIMIM fact with the party’s superior votes‑to‑seats conversion.

Minimal diagnostic that replicates the story

1. Seat–vote gap: report seat share minus vote share by party. (TRS ≈ +1.5 pp; BJP ≈ −3.6 pp; AIMIM ≈ +10.6 pp.)
2. Votes per seat: total party votes ÷ seats won (table above).
3. Margin distribution: sort wards by (winner − runner-up) margin; the top-tail is AIMIM-dominant.
4. Wasted‑vote rate: compute $\frac{W_p}{\sum_w v_{p,w}}$ using ward‑level counts.
5. Spatial concentration: a simple Gini over ward‑level party vote shares (or Moran’s I) will show AIMIM’s concentration relative to TRS/BJP.

What the pattern means

Residential and partisan segregation acts like a natural gerrymander. It doesn’t just decide who wins; it determines how efficiently parties harvest seats from the same citywide vote mass. In GHMC‑2020 that logic yielded: (i) an AIMIM seat bonus well above its vote share, (ii) extremely safe AIMIM wards with large surpluses, and (iii) many narrowly decided TRS–BJP contests that bled votes into losses.

Context

Rodden shows that in single‑member plurality systems (like Hyderabad municipal council), a party whose voters are packed into dense urban areas is often under‑represented city‑ or nation‑wide: it racks up a few hyper‑safe seats while the rival wins many districts narrowly. The margin distribution we see in GHMC—AIMIM owning the top‑20 margins—fits that packing signature.

So why does AIMIM still convert votes to seats efficiently here? Two facts break the simple “cities lose” intuition:

1. AIMIM is a sectional/territorial party within Hyderabad, not the largest citywide vote‑getter. Its cluster in the Old City spans many wards (not just a handful), producing a block of seats. Outside the cluster, AIMIM’s vote share is very low—so its losing‑ward waste is small.
2. TRS and BJP have diffuse support across the rest of the city and face each other everywhere, generating many close losses—large losing‑ward waste—plus narrow wins that still require substantial vote investment.

Heuristic: concentration hurts when your cluster is too small relative to district granularity (you get a few blowouts and little else). Concentration helps when the cluster covers enough districts to yield many wins, and your outside‑cluster share is near zero, keeping wasted votes in losses small. GHMC‑2020 is in the latter regime for AIMIM: hyper‑safe wins (top‑tail margins) and low losses elsewhere, hence a seat bonus.

Replication notes: Use the ward table to compute margins and wasted votes; the city has 150 wards, so seat share = seats/150. A proportional‑representation counterfactual (allocating 150 seats by vote share) provides a clean baseline for quantifying the FPTP seat bonus.