Why momo?

The Economist's Big Mac Index compares one burger across countries. Kathmandu runs on momo instead: every neighbourhood sells it, from a Rs 120 plate at a campus stall to a Rs 400 rooftop special. Tracking one dish everywhere gives a rough, tasty gauge of where the valley is cheap, where it's pricey, and how prices move over time.

This is a hobby project, not official statistics. Treat it like a conversation starter, not a price list.

1. Price per piece

Plates don't all have the same number of pieces: 8 here, 10 there, 12 at a lunch counter. So every report is divided into a price per piece: Rs 160 for 12 pieces is Rs 13.3 per piece.

The headline "average plate" is that per-piece figure × 10, a standard 10-piece plate, so every number on the site is comparable.

2. Median, not mean

Wherever the index summarises reports it uses a median: line the prices up and take the middle one. One fancy hotel plate can't drag the whole city up.

3. Why a model, not a raw average

Published momo prices are lumpy. In 2013 the only dated prices online are a Rs 150 plate on Durbarmarg and a Rs 400 hotel serving; in 2011 they're mostly Rs 30–40 old-town stalls. Averaging samples like that makes prices jump up and down from year to year. It even made veg look dearer than chicken, just because the veg reports happened to come from tourist cafés.

So every number on the site is an estimate of a typical plate:

typical plate = level × CPI × area × filling × style

4. The parts of the model

• CPI. Nepal's consumer price index (World Bank, 2010 = 100), carried into 2026 with the latest inflation rate (6.3%). The two big snapshots of published prices, adjusted for area, filling and style, show momo rising ×2.41 between early 2011 (21 prices) and 2025 (15 prices), while the CPI rose ×2.42. Momo has kept pace with inflation, so the CPI fills in the years between the snapshots and before 2011. For context, buff meat went from about Rs 200 a kilo in 2010 to Rs 500–600 now, and the minimum wage from Rs 6,200 a month (2011) to Rs 19,550 (2025).
• Level. Rupees per CPI point, fitted as the median over all published prices.
• Area. Each area starts from a multiplier based on rents, tourist trade and café density (Thamel 1.35×, Kirtipur 0.82×), then moves toward what its own reports say. With n reports it moves n/(n+8) of the way, so one odd plate can't redraw the map. Areas with no reports keep the starting estimate and are shown faded.
• Filling and style. Fixed mark-ups from comparing dishes on the same menu: veg 0.88×, buff 1×, pork 1.1×, and chicken 1.25× in 2006 falling to about 1.08× now, as chicken got cheaper relative to buff. Styles: steam 1×, jhol 1.12×, kothey 1.18×, fried 1.2×, C-momo 1.3×. “All types” and “All styles” use a rough sales mix.
• Outliers. Plates below 0.4× or above 2.5× of typical (like the Rs 400 hotel serving) stay in the table, marked “outlier”, but don't shape the model.

An area's index is its typical price divided by the city's: 1.4× means 40% above a typical Kathmandu plate. Areas need 3 or more reports to be named cheapest or priciest.

5. What your reports do

New reports pull the estimate for their period and area toward themselves the same way: a handful barely move it, dozens take over. In All prices, “vs typical” shows how each plate compares with a typical one for its area, filling, style and year. Sort by it to find the best deals.

6. Where the data comes from

History (2011–2026): prices printed in news articles, restaurant reviews and food blogs: The Kathmandu Post, Nepali Times, Republica, The Rising Nepal, Spotlight Nepal and several travel blogs. Every one of these rows is labelled “News / blog” and links to its source in All prices.

How those rows were recorded:

• Where a source didn't say how many pieces came on the plate, 10 is assumed (the usual Kathmandu plate). The table marks these with an asterisk, like 10 pcs*.
• Where a source gave a menu range (for example Rs 80–150), the midpoint is used and the range is kept in the note.
• Where a source didn't name the filling or style, it is filed as “Other”.
• One 2011 article is undated online; it is placed in early 2011 because it previews that year's Momo Mania festival.
• Localities are grouped into 26 areas (Jhochhen into Basantapur, Kamalpokhari into Naxal, and so on); the exact place is kept in the note.

Gaps: no dated online prices were found for 2006–2010, 2015–2018 or 2021. For those years the trend shows the model's estimate only, with no report dots.

New reports come from anyone. Each says whether it came from a visit, a menu, a delivery app (often marked up) or a news story. They're checked on the server: Rs 50–1,000 a plate, 5–15 pieces, and seen in the last 12 months.

Privacy

No accounts, no cookies, no tracking. To slow down spam, each report stores a one-way, salted hash of the sender's IP address. It's used only to limit how many reports one connection can send per hour, and it's never shown or shared. This phone keeps a copy of the price list so the site opens fast next time.