The Math That Decides Who Gets a Seat in Nepal's Parliament

You can't have 0.72 of a parliamentary seat.

That sounds obvious, but it's actually the core problem at the heart of every proportional representation system on Earth. When a party wins 35.2% of the vote and there are 110 seats to hand out, simple multiplication gives you 38.72 seats. That fraction — that stubborn remainder that won't resolve into a whole human being sitting in a chair in the legislature — is where the real politics of electoral math begins.

Nepal's answer to this problem is a formula invented by a French mathematician in 1910. It's called the Sainte-Laguë method, and it quietly determines the composition of roughly 40% of Nepal's House of Representatives. Most voters have never heard of it. But in a parliament where the effective number of parties hit 4.75 in 2022 — the most fragmented in Nepal's democratic history — the difference between one allocation method and another can determine who forms a government.

Here's how it works, why Nepal chose it, and what it means for the country's politics.

Nepal's parliament is actually two elections in one

Before we get to the math, some quick context. Nepal doesn't run a single election. It runs two, simultaneously, on the same day.

The first is straightforward: 165 constituencies, each electing one representative through First-Past-The-Post. Whoever gets the most votes wins. Simple. Brutal. Sometimes wildly disproportionate.

The second is the PR election. Every voter casts a separate ballot for a political party (not a candidate), and those votes get converted into 110 seats using the Sainte-Laguë method. The PR seats exist specifically to correct the disproportionalities that FPTP creates.

Together, the 275-seat House of Representatives is supposed to balance local representation with national proportionality. That's the theory, anyway. Whether it works depends a lot on the math.

The formula is simpler than you'd expect

The Sainte-Laguë method allocates seats one at a time. Each round, every party gets a score — a quotient — calculated by dividing their total votes by a number that increases as they win seats.

The formula:

Quotient = Votes ÷ (2s + 1)

Where s is the number of seats that party has already been allocated.

So a party starting with zero seats divides its votes by 1. After winning its first seat, it divides by 3. After its second, by 5. Then 7, 9, 11, and so on. The divisor sequence is just odd numbers: 1, 3, 5, 7, 9, 11...

Each round, the party with the highest quotient wins the next seat. Their quotient drops. Everyone else stays the same. Repeat 110 times. Done.

That's it. That's the whole method.

Let's actually do the math

Abstract formulas are boring. Let's allocate some seats.

Three parties. Ten seats. Here are the votes:

• Party A: 10,000 votes (50%)
• Party B: 6,000 votes (30%)
• Party C: 4,000 votes (20%)

Round 1: Everyone starts at zero seats, so everyone divides by 1.

Party A wins Seat 1. Its quotient now gets recalculated.

Round 2: Party A divides by 3 now.

Party B wins Seat 2. Notice what happened: Party A had 10,000 votes but its quotient dropped below Party B's. The method is self-correcting.

Round 3:

Party C wins Seat 3. Every party now has one seat.

Round 4:

Party A wins again. And the process continues — each time, the leading party's quotient drops, letting others catch up.

Final result after 10 rounds:

• Party A: 5 seats (50% of votes → 50% of seats)
• Party B: 3 seats (30% of votes → 30% of seats)
• Party C: 2 seats (20% of votes → 20% of seats)

Perfect proportionality. In this case, at least.

Why odd numbers? What's special about 1, 3, 5, 7?

Here's the intuition: each new seat a party wins makes the next one harder to get. The divisor jumps by 2 each time, which means a party's quotient drops steeply after each win. That steep drop is what prevents big parties from hoarding seats in the early rounds.

Compare this to the D'Hondt method, which is the other major allocation system used around the world. D'Hondt uses divisors of 1, 2, 3, 4, 5, 6... The gaps are smaller, which means a large party's quotient stays higher for longer. The result? D'Hondt systematically gives a slight bonus to bigger parties.

The difference isn't academic. It changes who sits in parliament.

What the 2022 election actually looked like

Let's apply this to real data. In the 2022 election, here's how the PR votes broke down among parties that crossed the 3% threshold (approximate vote totals derived from official Election Commission percentages and a total of ~10.56 million valid PR ballots):

UML led NC by roughly 131,000 votes — a margin of 1.24 percentage points. That gap translated into a two-seat advantage in PR allocation (34 seats vs. 32).

Here's how Sainte-Laguë allocated the 110 seats:

The average deviation between vote share and seat share? About 1.7 percentage points. That's good — not perfect, but far better than pure FPTP systems, which routinely produce deviations of 10–15 percentage points.

Notice that every qualifying party is slightly overrepresented relative to their raw vote share. This is because roughly 12% of total PR votes went to parties that fell below the 3% threshold and received no seats — those votes effectively inflate the seat shares of the parties that did qualify.

With Sainte-Laguë, the overrepresentation is distributed roughly in proportion to party size rather than being concentrated among the largest parties. And that matters.

What if Nepal used D'Hondt instead?

Run the same 2022 votes through D'Hondt and you get a slightly different parliament:

One seat shifts from the smallest qualifying party (JP, with 3.74% of votes) to the largest (UML, with 26.95%). That's D'Hondt doing what D'Hondt does: giving a small bonus to big parties at the expense of small ones.

One seat doesn't sound like much. But in a parliament where the effective number of parties is 4.75 and coalition math is everything, a single seat can be the difference between a viable coalition and a failed one. In 2022, the eventual governing coalition had a working majority of just a handful of seats. Move one here, one there, and the whole calculus shifts.

The 3% threshold: democracy's bouncer

Before Sainte-Laguë even kicks in, there's a gate to pass through. Parties need at least 3% of the national PR vote to qualify for seats.

In 2022, parties that fell below 3% collectively received roughly 1.3 million PR votes — about 12% of all valid PR ballots cast. That includes the Unified Socialists (2.83%), the National Unity Party (2.57%), the Loktantrik Samajbadi Party (1.58%), and numerous smaller parties. Those votes just... disappeared. They didn't count toward any seat allocation. The voters who cast them got FPTP representation in their local constituencies, but their PR votes were effectively wasted.

This creates an interesting tension. The whole point of PR is to ensure every vote counts. But the threshold means some votes don't. The justification is practical: without a threshold, you'd get a parliament full of micro-parties with one or two seats each, making coalition-building nearly impossible.

Is 3% the right number? Germany uses 5%. Israel uses 3.25%. Turkey uses 7% (which has been widely criticized as too high). Nepal's 3% is on the lower end internationally, which means it's relatively permissive. In 2022, seven parties cleared it. That's a lot of parties — but not an unmanageable number.

The threshold also creates strategic incentives. If you're a voter whose preferred party is polling at 2.5%, you face a dilemma: vote your conscience and risk wasting your ballot, or vote strategically for a larger party you're less enthusiastic about. This is the same logic that drives third-party anxiety in American elections, just with a different mechanism.

How PR corrects FPTP's distortions

Here's where the two halves of Nepal's electoral system interact in interesting ways.

In 2022, the FPTP results alone gave NC 57 seats, UML 44, and Maoist Centre 18 out of 165. Those numbers don't look too far off from the parties' national vote shares — but that's partly coincidence. FPTP can produce wildly disproportionate outcomes depending on how votes are distributed geographically.

The PR seats act as a corrective. UML picked up 34 additional seats through PR, bringing its total to 78 (28.4% of the 275-seat house). NC got 32 more, reaching 89 (32.4%). And critically, parties like RSP — which won 20 total seats, with 13 coming through PR — got parliamentary representation that FPTP alone would have denied them.

RSP's story is actually the best argument for why PR matters. The party didn't exist before 2022. It emerged as a genuine grassroots force, winning 10.70% of the PR vote and 7 FPTP seats for a total of 20 seats. Under a pure FPTP system, a new party with geographically dispersed support would struggle to win individual constituencies. PR gave RSP a voice broadly proportional to its actual support. Whether you like RSP's politics or not, that's democracy working as designed.

The gender quota adds another layer

Nepal's PR system doesn't just allocate seats to parties. It also enforces gender representation. Each party's PR candidate list must alternate between men and women — at least one woman for every two candidates.

So when UML wins 34 PR seats, those seats go to the first 34 candidates on UML's pre-submitted list. And because the list alternates by gender, at least 17 of those 34 seats go to women.

This is a separate mechanism from Sainte-Laguë itself. The allocation formula determines how many seats each party gets. The candidate list determines who fills those seats. But the two systems work together to produce a parliament that's both proportionally representative of party support and more gender-balanced than FPTP alone would deliver.

Edge cases: what happens when things get weird

Ties. What if two parties have identical quotients when the last seat is being allocated? It's theoretically possible, though rare with millions of votes in play. Nepal's rule is simple: draw lots. Flip a coin, essentially. In 2022, it didn't come up. But in a close election with smaller vote totals, it could.

Can you end up with 109 or 111 seats? No. This is one of Sainte-Laguë's nice properties. Because seats are allocated one at a time — stop at exactly 110 — you always get exactly the right number. Some other methods, like largest remainder systems, can require awkward adjustments. Sainte-Laguë doesn't have that problem.

What about the Alabama Paradox? This is a famous quirk in apportionment math where increasing the total number of seats can actually decrease a party's allocation. It's theoretically possible with Sainte-Laguë but essentially never happens when you're allocating 100+ seats. Nepal's 110 PR seats are well within the safe zone.

The mathematical properties (for the nerds)

I'll keep this brief, but Sainte-Laguë has some properties that mathematicians find satisfying.

Monotonicity: If a party gains votes while everyone else stays the same, it can't lose seats. Sounds obvious, but not all allocation methods guarantee this.

House monotonicity: If you increase the total number of seats, no party loses seats. Again, not guaranteed by every method.

Minimum deviation: The Balinski-Young theorem shows that Sainte-Laguë minimizes the maximum deviation from exact proportionality. In plain English: of all the divisor methods, it gets closest to giving every party exactly the seat share their vote share deserves.

That last property is the big one. It's why Sainte-Laguë is generally considered the fairest divisor method available. Not the simplest. Not the most intuitive. The fairest.

Who else uses this system?

Nepal isn't alone. New Zealand uses Sainte-Laguë for its MMP system (which is structurally similar to Nepal's mixed system). Germany uses it for certain calculations. Norway and Sweden use a modified version where the first divisor is 1.4 instead of 1, which slightly penalizes very small parties.

On the other side, most of Europe and Latin America use D'Hondt. Spain, Belgium, Switzerland, Portugal — they all give that small bonus to larger parties. The choice between the two methods isn't random. It reflects a country's priorities. Do you want maximum proportionality, even if it means more parties in parliament? Or do you want a slight tilt toward governability?

Nepal chose proportionality. Given that the country's parliament already had an effective number of parties of 4.75 in 2022 — and that coalition politics has become the norm rather than the exception — that choice has real consequences. More parties get seats. Coalitions get more complicated. But the parliament more accurately reflects what voters actually want.

The misconceptions that won't die

"It's too complicated." The formula is division. Your phone can do it. Voters don't need to understand the algorithm any more than they need to understand how their car's engine works. They just need to trust that it converts their votes into seats fairly. And the data shows it does.

"PR creates too many small parties." The 3% threshold exists precisely to prevent this. Seven or eight parties in parliament isn't fragmentation — it's pluralism. Germany has a similar number. So does New Zealand. These are functional democracies.

"FPTP is more democratic because you know your representative." This one has some merit — there is value in having a local representative you can hold accountable. That's why Nepal kept 165 FPTP seats. But FPTP alone can give a party with 38% of the vote a 53% majority in parliament. That's not democratic either. The mixed system tries to get the best of both.

What we know and what we don't

Here's what the data tells us clearly: Sainte-Laguë does what it's supposed to do. In 2022, the average deviation between vote share and seat share was about 1.7 percentage points. That's about as good as you can get when converting percentages into whole numbers.

We also know the system has shaped Nepal's party dynamics. The combination of PR seats and a low threshold gave RSP a meaningful parliamentary presence in its very first election. It gave smaller parties like RPP and JSPN seats roughly proportional to their support. And it prevented any single party from translating a plurality of votes into an outright majority.

What we don't know is harder to pin down. Does the proportionality of Sainte-Laguë make coalition politics more stable or less? Nepal has had plenty of unstable coalitions, but it's hard to separate the effect of the allocation method from the effect of, well, everything else going on in Nepali politics. The 2022 parliament was the most fragmented in the country's history. Was that because of Sainte-Laguë, or because voters genuinely wanted more choices? Probably both, in proportions we can't precisely measure.

There's also an open question about whether the 3% threshold is set correctly. Lower it and you get more representation but more fragmentation. Raise it and you get cleaner coalition math but more wasted votes. Nepal's current threshold seems to be in a reasonable range, but "reasonable" is a judgment call, not a mathematical certainty.

The formula itself — Votes ÷ (2s + 1) — is elegant in the way that good math often is. It solves a genuinely hard problem with a genuinely simple mechanism. But no formula, however elegant, can solve the political problems that come after the seats are allocated. That part is still up to the politicians.

And if Nepal's recent history is any guide, that's where things get interesting.

All calculations use official Election Commission of Nepal data. For an interactive Sainte-Laguë calculator, visit our simulator.