Game Theory & Nepal's Elections

March 5, 2025

I know very little game theory and I know very little politics. I find both greatly interesting.

Game theory is the study of interactions between rational (?) decision-makers. Rational decision-makers. Game theory assumes rationality.

Out of many game theory concepts applicable in elections, there are a few I find relevant to Nepal's system (multi-party democracy with FPTP and PR).

The median voter theorem. Notice how every single candidate has the same lukewarm take. Zero tolerance to corruption, and good governance. Closely connected is the Spatial Competition Model (Downsian Model) — policies positioned to maximize votes (e.g. employment, governance, anti-corruption). This is the main reason we see generic policies (instead of extremely detailed).

Connected Coalition Theory. Coalitions among contiguous ideological actors (for example, center joins center-left bloc). This is relevant to NCP, Maoist Center merging with other left-wing forces.

Strategic Voting (Tactical Voting). Voters picking the viable candidate over their preferred longshot. Rather support a second choice to block a rival. An example would be core NC supporters voting for RSP in areas with NC candidates they perceive as weak instead of voting for other leftist parties (support second choice instead of staying loyal to avoid the third).

Entry Deterrence Game. It's a natural law. People avoid playing games they think are not likely to win.

Shapley–Shubik Power Index. Measures the probability of party being decisive vote. Small parties become coalition kingmakers. RSP being part of the cabinet with just 20 seats is an example. Pushpa Kamal Dahal "Prachanda" being the prime minister on occasions is also an example of this.

Threshold Voting Dynamics. Voters avoid parties near the exclusion threshold. Small parties disappear.

Colonel Blotto Strategy. Campaign resources allocated across competitive districts and spend heavily in swing seats (obviously).

Bandwagon Effect. Voters support the candidate perceived as likely winner. Poll surge increases support. Rallies and mass gatherings create this whim. Camera angles positioned to show a larger crowd are a key tactic.

Coordination Game Among Voters. Voters converge on viable candidates collectively. Polls push voters to two leaders.

Some might comment on my exclusion of Duverger's Law. I read a blog published by LSE on this. This law says that FPTP voting results in two-party systems. The blog argues that Duverger's Law is extinct. Nepal's history dictates the same — Prachanda becoming the prime minister while his party finished third is a prime example.

A "Game Theoretic" Breakdown of Morang-6

I vote at Morang constituency 6. Here is a "game theoretic" breakdown. A disclaimer to readers — a) I am writing based on public domain information, and b) there perhaps are better constituencies to analyse (Sarlahi 4 because of the Gagan-Amresh-Shiva Pujan trifecta or Jhapa-5 for the Oli-Balen tug-of-war).

Twenty candidates filed nominations here. Key figures — Dr. Sekhar Koirala (NC), Binod Dhakal (CPN-UML), Opendra Raya (NCP, Maoist center), Rubina Acharya (RSP).

The early deterrence game and the bandwagon

Candidates from Nagarik Unmukti Party, Janata Samajwadi Party, Janmat Party, Nepal Workers and Peasants Party, Janta Samajwadi Party and CPN, Unified Socialist withdrew their nomination.

There was a rumor saying Basanta Budhathoki (Ujyalo Nepal Party) withdrew declaring support for Binod Dhakal. Later, the Ujyalo Nepal Party issued a press release claiming it was not the case.

All candidates who withdrew unanimously supported Dr. Sekhar Koirala.

Vote fragmentation

Days before Dr. Sekhar Koirala inducted 150 people (allegedly formerly Maoists, UML and other) to Nepali Congress — a key tactic to circumvent vote fragmentation. The candidates who withdrew supporting him was also another tactic for the same.

The bandwagon effect and the coordination game

This is arguably the most important factor in this election. The Balen Bandwagon is a force to be reckoned with. Posters in this area with the face of Balen instead of the candidate speak volumes. Many in the area say they are voting for Balen, most did not know the name of Rubina Acharya. Coordinated voters across the nation are voting for candidates endorsed by RSP are doing so for Balen, not for RSP, and not for the candidates as an individual (a little concerning for democracy imo, and obviously does not apply evenly to all constituencies).

Prisoner's Dilemma and the Duvergerian Squeeze

The Prisoner's dilemma facing anti-UML voters is resolved in this race by three factors: a) Dr. Sekhar Koirala's incumbency (a coordination anchor), b) his active absorption of smaller party networks, and c) the national focal point of the 2025 protests. Rational anti-UML voters who prefer RSP but fear a repeat of UML winning on a split vote face a classic Duvergerian squeeze: the expected-value calculation favours tactical migration toward the most viable anti-UML candidate, which is Koirala.

RSP supporters with strong anti-establishment preferences face the harder calculation. If their sincere preference is RSP but they assign positive value to preventing UML from winning, the dominant strategy depends on their estimate of RSP's viability. Given that Acharya is a first-time candidate with no local organisational depth against Dr. Sekhar Koirala's two-election incumbency, most rational RSP-leaning voters will ultimately defect to NC on the FPTP ballot — while expressing sincere RSP preference on their PR ballot.

The reality

Morang-6 is a competitive three-way race. The coordination problem for Anti-NC and Anti-RSP voters is easy compared to Anti-UML voters. The coordination problem for Anti-UML voters is difficult — split centrist and/or right-wing votes between NC and RSP, leading to a UML victory (after all this)?

We have three scenarios:

A. Dr. Sekhar Koirala's organisational depth, incumbency coordination advantage, and pre-election consolidation hold. Acharya draws heavily from UML defectors and first-time voters but cannot overcome Dr. Sekhar Koirala's established ward-level machinery. The wave registers as a strong RSP second or close third, but FPTP's winner-take-all structure rewards Dr. Koirala's plurality. This was the base case; it remains the modal outcome, but no longer comfortably so.

B. RSP wins on wave momentum. This is now a non-trivial possibility rather than an upset scenario. The conditions for Scenario B are: first-time voter turnout is disproportionately high in Biratnagar's urban wards; Acharya runs a visible, energetic local campaign that successfully anchors the national Balen enthusiasm to her candidacy; and a meaningful share of soft NC supporters — particularly younger ones — decide that this election is the moment to vote sincerely for RSP rather than tactically for the incumbent. If these conditions hold simultaneously, Acharya's vote share could exceed Dr. Koirala's in the urban wards and partially offset his rural and organisational advantage.

C. UML wins on a three-way split. If the NC-RSP split is severe enough — say, 30% each — UML's consolidated base of 32–35% produces a plurality victory for Dhakal. This is the classic Duvergerian nightmare: two anti-UML candidates failing to coordinate produces the worst collective outcome for both their supporters. The game-theoretic irony is that the stronger the ghanti wave, the more it risks producing a UML win in this specific constituency through vote fragmentation.

In summary, it remains unclear who will win here.