How to Size a Market in Nepal: Sources That Hold Up

Ask five people how big a district market in Nepal is and you will get five numbers, each delivered with total confidence, none of them showing the arithmetic. The confidence is usually the problem. Most district figures floating around sit on top of an assumption nobody printed, and when that assumption breaks, the number goes with it.

This guide is about the plumbing underneath the number. Which datasets actually measure demand, which ones only look like they do, and how to combine them in a sequence you can defend when a client, a lender or a sceptical colleague asks where the figure came from.

Start by naming the thing you want to measure

Market size can mean at least four different quantities: how many households exist, how much those households spend, how many businesses operate, or how much money moves through the formal economy. In Nepal, the data behind each one comes from a different institution, on a different cycle, at a different geographic resolution. Mixing them without noticing is where estimates quietly fall apart.

Market sizing in Nepal: four data families, four different jobs

Census household counts

The national census is the best available source for the denominator. It tells you how many households and people live where, down to very fine geography. What it does not tell you is what any of them spend. A household count is a multiplier waiting for an assumption, not an answer.

Household consumption surveys

These surveys measure spending, income and living standards in real detail, but the sample is built to represent the country and its broad regions. It is not built to represent one district. Borrowing a national average consumption figure for a single district is an assumption, and a fairly aggressive one, because spending patterns shift with remoteness, farming, remittances and urbanisation. If you do it, say so on the page.

Economic census establishment counts

An establishment count tells you how many fixed places of business exist in an area and roughly how many people they employ. That is genuinely useful for thinking about supply and competition. It is a weak proxy for market value on its own, because an establishment is not a firm, and published counts do not carry turnover or revenue.

Central bank statistics

Central bank publications give you national accounts, balance of payments, credit and price aggregates. They are the right place to test whether your district number is plausible against national totals. They will not hand you district demand, and no amount of interpolation turns a national aggregate into a local fact.

The combination logic, in four steps

Take the household count for the area from the census. Choose a spending assumption you can justify, and write down the justification. Multiply. Then present a range rather than a point estimate, with every input labelled by source and date.

That last step is not decoration. A district-level demand estimate for Nepal is a model, and a model is only as good as the assumptions printed beside it. Most Nepal market research stops at the headline figure and skips this entirely, which is exactly why so many headline figures cannot be checked.

A worked example, with numbers I made up

Illustrative only. Every figure below is invented to show the arithmetic. None of it describes a real district, and none of it should be quoted as data.

Numbers invented for illustration. Replace this table with the published household and population counts from the census table you are citing, and keep the table number, the release date and the date you retrieved it alongside it.

Now the assumption. Suppose each household spends NPR 1,800 a month on the category you care about. Annual spend per household is 1,800 x 12 = 21,600. Multiply by households: 7,035 x 21,600 = NPR 151,956,000, or roughly NPR 152 million a year.

That single number is not the deliverable. The range is. Flex the spending assumption by 20 percent either way and you get NPR 122 million to NPR 182 million. The household count is solid; the spending assumption is doing most of the work, and the range is an honest admission of that.

Caveats that survive every dataset

Household surveys are nationally and regionally representative, not district representative. Establishment counts measure premises, not revenue. Central bank aggregates are national by construction. Census counts give you denominators, never spending. Any district figure built from these is a denominator plus an assumption stack, and hiding the stack behind one confident number is the thing to avoid.

The figures that go stale

Before you publish, build a small register of every perishable item and attach it to whoever owns it, plus the date it was issued. At minimum: the census table number and its release date, the census reference date printed on that table (it is not always the fieldwork year), the survey round you used and its fieldwork years, the economic census round and whether a newer one exists, the exact title and issue of any central bank publication, and the count of local units if you plan to mention one. Reconfirm each line against the live source on the day you publish. If a line cannot be reconfirmed, cut it rather than ship a stale date.

One more check that people skip: confirm the name of the body currently publishing the statistics you cite. Statistical agencies get restructured, and a citation to an office that has since been renamed or absorbed undermines the whole piece even when the number itself was right.

What to do with all this

Use the four families for the jobs they were built for. Print the assumption. Show the range. Date every source. A district estimate with visible arithmetic is more persuasive than a confident single figure, because anyone reading it can see exactly where to push back, and that is what makes it usable.