You Can't Win Nepal Without Winning Madhesh

Forget Kathmandu. If you want to predict who'll run Nepal's next government, drive south. Past the hills, past the Chure range, into the flatlands that stretch toward the Indian border. That's where elections are won and lost.

Madhesh Province has 32 parliamentary constituencies — 19 percent of the 275-seat House of Representatives. On paper, that doesn't sound like a kingmaker. But here's the thing: no party since federalization has won a majority without locking down at least 12 to 15 of those seats. And the region's internal politics are volatile enough that a 20-seat swing between elections isn't just possible — it's happened.

Between 2017 and 2022, regional parties went from holding 24 of Madhesh's 32 seats to holding just 12. National parties flipped the script almost entirely. That kind of realignment in a single election cycle is extraordinary. And it raises an obvious question heading into 2027: is the shift permanent, or was it a blip?

The answer depends on who you ask. But the data points in one direction more than the other.

The math is simple, and it's brutal

Let's start with arithmetic. You need 138 seats for a majority. The three big national parties — Nepali Congress, UML, and Maoist Centre — fight over roughly 200 constituencies across the rest of the country. In those races, they tend to split the vote three ways, with smaller parties grabbing scraps. Nobody comes out of those 200-odd seats with a commanding lead.

That leaves Madhesh as the tiebreaker. Win big here, and you're forming a government. Lose badly, and you're negotiating from weakness.

Look at how the 2022 results broke down in the province:

Figure: Seat distribution in Madhesh Province (2022)

Five different party families won seats. That fragmentation is the story. In 2017, you could draw a clean line: regional parties dominated, national parties were marginal. By 2022, the picture had shattered into pieces. And shattered pictures create opportunities — for everyone.

NC's 8 seats in Madhesh were arguably more important to its governing math than any 8 seats it won elsewhere. Without them, the post-election coalition arithmetic looks completely different. KP Sharma Oli knows this too. UML's 3 seats in Madhesh were embarrassing, and the party has been quietly investing in the region ever since.

What happened between 2017 and 2022

The 2017 election was supposed to be the dawn of Madheshi political power. Regional parties — primarily the RJPN and the Federal Socialist Forum — had fought hard for federalism, protested the 2015 Constitution's provincial boundaries, and mobilized massive support. They swept 24 of 32 seats. It was a mandate.

Five years later, they held 12.

So what went wrong? A few things, and they compounded each other.

The merger that wasn't. RJPN and the Federal Socialist Forum merged to create the Janata Samajbadi Party Nepal (JSPN). On paper, consolidation should have strengthened them. In practice, it created internal power struggles that never resolved. Splinter groups broke off. Candidates defected. The organizational machinery that won 2017 rusted out.

National parties got smarter. NC and Maoist Centre didn't just run candidates in Madhesh — they recruited credible Madheshi leaders. This wasn't tokenism. They gave these leaders real positions, real resources, and real autonomy. Voters noticed.

Young voters checked out of identity politics. About 30 percent of Madhesh's electorate is under 35. These voters grew up after the Madhesh movement. They care about jobs, infrastructure, and economic opportunity. "Madheshi identity" isn't irrelevant to them, but it's not sufficient either. When regional parties ran on identity alone without a credible development agenda, younger voters looked elsewhere.

Economic pragmatism won. This one's related to the youth shift but broader. Madhesh has the highest poverty rate in Nepal — roughly 27 percent, compared to the national average around 19 percent. But it's also been growing fast, around 6.8 percent annually between 2018 and 2023. Voters in a region that's poor but getting richer fast tend to reward parties that can credibly claim they'll keep the growth going. National parties, with their access to federal budgets, had an easier time making that case.

The three identities every candidate has to manage

Here's where Madhesh gets complicated — and where a lot of outside analysts get it wrong.

Madheshi voters don't think about politics through a single lens. They're juggling at least three overlapping identities, and the weight each one carries shifts depending on the election, the candidate, and the issue.

First, regional identity. Madhesh has a distinct cultural and linguistic character. Hindi, Maithili, Bhojpuri, and other languages dominate daily life. There's a long history of feeling marginalized by hill-centric governance in Kathmandu. The 2015 border blockade — when Madhesh-based protests coincided with an Indian economic blockade that crippled the country — crystallized this sense of separateness. Regional identity still matters. A lot.

Second, national identity. But Madheshi voters are also Nepali citizens who want federal benefits, national infrastructure investment, and a seat at the table in Kathmandu. The 2015 Constitution, for all its flaws from a Madheshi perspective, created a federal structure that gives the province real governing power. Voters increasingly want leaders who can work the national system, not just protest against it.

Third, sub-regional identity. This is the one outsiders miss most often. Madhesh isn't monolithic. Caste, district, and personal loyalty to local leaders matter enormously. A Yadav candidate in Dhanusha faces different dynamics than a Muslim candidate in Rautahat or a Tharu candidate in the western districts. Party labels sometimes matter less than the candidate's family name and local reputation.

The party that threads all three needles wins. In 2017, regional parties won by emphasizing the first identity at the expense of the other two. By 2022, voters wanted more balance.

The caste math that nobody can ignore

I need to talk about caste, because you can't understand Madhesh elections without it.

The region's demographics break down roughly like this:

Dominant caste groups — Yadav, Tharu, Muslim communities, and others — make up about 65 percent of the population. They traditionally backed regional parties but have been fragmenting across multiple parties since 2022. This is the key swing demographic. Whichever party consolidates support here has an enormous advantage.

Upper caste groups — Brahmin, Kayastha, Rajput — are about 20 percent. They've historically leaned toward national parties, and their influence exceeds their numbers because they tend to provide organizational capacity: the local lawyers, teachers, and businesspeople who run campaign operations.

Marginalized communities — Dalit groups and other extremely marginalized populations — make up roughly 15 percent. They're becoming more politically assertive, demanding representation within whatever party they support. In tight races, their votes are decisive.

Here's the bottom line: no single caste group can deliver a constituency on its own. Every winning candidate needs a multi-caste coalition. The days when a dominant-caste leader could simply mobilize their community and coast to victory are over. That's actually a sign of democratic maturation, even if it makes elections messier and harder to predict.

Development is the new battleground

For years, Madhesh elections were about identity and autonomy. Those issues haven't disappeared. But they've been joined — and in some constituencies overtaken — by bread-and-butter development questions.

The numbers tell a contradictory story. Madhesh is simultaneously Nepal's poorest province and one of its fastest-growing. Poverty sits at 27.4 percent, well above the national average of 18.7 percent. But look at the development indicators from the past five years:

• Road connectivity up 47 percent
• Electrification at 89 percent, up from 52 percent in 2017
• The Birgunj-Bardibas industrial corridor is booming
• Remittance inflows per capita are the highest in Nepal

So things are getting better. But inequality is rising too. The Gini coefficient is 0.38 — high by South Asian standards. The urban-rural gap is widening. Youth unemployment sits at 16.7 percent, above the national average.

What does this mean politically? Voters are economically literate and demanding. They've seen what development looks like — new roads, electricity, industrial jobs in the corridor towns — and they want more of it. Generic promises about "prosperity" don't cut it anymore. Parties need to show concrete results: this road, that factory, these jobs.

That's an advantage for incumbent parties at the provincial and federal level, assuming they've actually delivered. It's a disadvantage for opposition parties running on identity alone. And it's a wildcard for newer parties like RSP, which can promise a fresh approach without the baggage of past failures.

The India question (handle with care)

I'd be lying if I said you could analyze Madhesh politics without talking about India. The province shares more than 800 kilometers of border with Bihar and Uttar Pradesh. About 70 percent of Nepal's trade with India passes through Madhesh. Family connections cross the border freely. Indian media — Hindi news channels, Bollywood, cricket — is consumed as readily as Nepali media, maybe more so.

This creates a political minefield.

Any perception that a Madheshi party is too close to India backfires. The 2015 blockade poisoned the well for years. Voters remember empty gas cylinders and cold winters. They don't want to be seen as India's proxies.

But blocking Indian economic ties would be catastrophic. Madhesh's economy depends on cross-border trade, Indian investment, and the flow of goods through Birgunj and other border towns. Voters know this too.

So the winning formula is a kind of political judo: defend Nepali sovereignty loudly while maintaining beneficial economic ties quietly. Voters have gotten sophisticated about this. They can tell the difference between genuine nationalism and performative anti-India rhetoric. They reward leaders who manage the relationship pragmatically.

Recent trends suggest this sophistication is growing. Candidates who ran on simplistic "pro-India" or "anti-India" platforms in 2022 mostly lost. The winners were pragmatists who could hold both ideas in their heads at once.

Twelve constituencies that will decide everything

Our model identifies 12 hyper-competitive Madhesh constituencies that are likely to determine the shape of the next parliament. These are the seats where victory margins have historically been under 5 percent, where multiple parties have realistic shots, and where turnout swings can flip outcomes.

Tier 1: The five must-wins

Saptari-2, Dhanusha-3, Mahottari-2, Sarlahi-3, and Rautahat-2. These are the bellwethers. In every election since 2017, the party that won three or more of these five seats went on to lead the Madhesh coalition. They share common characteristics: diverse caste compositions, high voter turnout, competitive multi-party races, and thin margins. Dhanusha-3, for example, was decided by fewer than 2,000 votes in 2022.

Tier 2: The four coalition-builders

Bara-2, Parsa-3, Siraha-4, and Sunsari-2. Alliance dynamics matter more here than anywhere else. These constituencies tend toward three-way races, which means vote-splitting determines outcomes. A party running alone will lose. A party with the right local alliance partner can win. The youth vote is particularly decisive in this tier — these constituencies have younger-than-average electorates and higher rates of first-time voters.

Tier 3: The three wildcards

Udayapur-1, Saptari-3, and Mahottari-4. These are the constituencies where the old identity politics framework is most directly competing with the new development-first message. There's a visible generational divide: older voters lean toward regional parties, younger voters toward national parties or RSP. These seats could surprise everyone, and they're the hardest to model because the underlying political dynamics are shifting fastest.

What national parties need to do (and what they keep getting wrong)

Every election cycle, national party strategists fly into Madhesh, hold a few rallies, and fly back to Kathmandu convinced they've cracked the code. They usually haven't.

Here's what actually works, based on the last three election cycles:

Authentic local leadership. Parachute candidates — well-known national figures dropped into Madhesh constituencies — fail almost every time. Voters want leaders who live in the district, speak the local language natively, and have roots in the community. NC learned this the hard way in 2017 and adjusted by 2022. UML still hasn't fully absorbed the lesson.

Genuine commitment to provincial autonomy. Madhesh voters fought for federalism. They take it personally when national parties try to claw back provincial powers or starve provincial budgets. Any party perceived as centralizing will pay a price at the ballot box. This isn't abstract — voters track whether their provincial government has the budget to build roads and run hospitals.

Visible development delivery. Not promises. Actual things. A new bridge. A functioning health post. An industrial park that created jobs. Voters in Madhesh are pragmatic, and they have long memories. A party that built something tangible has a story to tell. A party that only talked about building things doesn't.

Cultural respect. This sounds soft, but it matters. Language policy, representation in national media, recognition of Madheshi festivals and cultural events — these are signals that a party sees Madhesh as an equal partner, not a vote bank to be harvested. Small symbolic gestures can shift perceptions more than big policy papers.

Multi-caste ticket balancing. Parties that nominate candidates from only one caste group across their Madhesh slate lose. Period. The ticket has to reflect the province's diversity. This requires discipline from party leadership, because local power brokers always push for their own community's candidates.

Looking ahead to 2027

So where does all this leave us for the next election?

Based on demographic trends and the trajectory of the past two cycles, here's our best estimate of how Madhesh's 32 seats might break down:

Figure: Projected seat distribution in Madhesh (2027)

The big question mark is whether regional parties can re-consolidate. The current trajectory points toward further fragmentation, which benefits national parties. But politics isn't physics — trends can reverse. A charismatic Madheshi leader who can reunify the regional vote would change the math overnight. Nobody fitting that description has appeared yet. That doesn't mean nobody will.

There's also the RSP factor. Rabi Lamichhane's party has been making inroads with young voters across Nepal, and Madhesh's under-35 population is large and restless. If RSP can build real organizational capacity in the province — not just social media buzz but actual ground-level infrastructure — it could pull 3 to 6 seats that would otherwise go to national parties. That's enough to scramble everyone's coalition math.

The caveats (because there are always caveats)

I want to be honest about what we don't know.

The projections above are based on historical election results and demographic trends rather than current polling, so they carry significant uncertainty. Madhesh is diverse enough that district-level projections are especially difficult to make with confidence. We're more confident about province-wide directional trends than individual constituency calls.

We're also projecting 18 months out. A lot can happen. A new political crisis, an economic shock, a major infrastructure project completion, or a dramatic party split could reshape the landscape entirely. Our projections assume rough continuity in current trends. That assumption will probably be wrong in at least a few constituencies.

And there's a measurement problem with caste-based voting patterns. Voters don't always tell pollsters the truth about caste considerations in their choices. Our demographic models try to account for this, but the correction is imperfect.

What we know, what we don't

Here's what the data tells us with reasonable confidence:

We know that Madhesh's 32 seats are disproportionately important to government formation. The math hasn't changed, and it won't change before 2027.

We know that regional parties lost ground between 2017 and 2022, and the factors driving that shift — youth demographics, economic pragmatism, national party recruitment of Madheshi leaders — haven't reversed.

We know that caste coalitions, development delivery, and the India question are the three axes around which Madhesh politics rotates. Any analysis that ignores one of these is incomplete.

We don't know whether regional party fragmentation will continue or whether a consolidation event is coming. The difference between those two scenarios is probably 8 to 10 seats — enough to determine who governs Nepal.

We don't know how RSP's entry will reshape the competitive dynamics. It could split the anti-establishment vote and help incumbents, or it could channel youth frustration into a coherent political force that displaces weaker parties.

And we don't know whether the development-first trend in voter preferences is durable or whether a new identity-based mobilization could reverse it. The conditions for such a mobilization — perceived disrespect from Kathmandu, a border incident with India, a rollback of provincial powers — exist. They just haven't materialized.

What we can say is this: anyone who tells you they know exactly how Madhesh will vote in 2027 is selling something. The region is too diverse, too dynamic, and too politically sophisticated for simple predictions. But anyone who ignores Madhesh in their national projections is making a bigger mistake.

The kingmaker is paying attention. The question is whether the parties are paying attention back.

Data sources: Election Commission Nepal (constituency results 2008–2022), Madhesh Provincial Government statistics, Central Bureau of Statistics (Census 2021), Nepal Rastra Bank economic data. Seat projections and demographic estimates carry significant uncertainty and should be treated as indicative ranges, not precise predictions. Economic figures (poverty rate, Gini coefficient, etc.) are drawn from the most recent available government and World Bank data and may not reflect conditions at the time of the next election.