The Nyra Domestic Demand Framework

Why a framework rather than a number

Most demand analysis in small, data-constrained markets fails the same way: it starts from a population figure, applies a per-capita spending assumption borrowed from somewhere else, and produces a market size with no error bars and no audit trail. The number is then used to underwrite an investment.

This framework exists to make that failure impossible to commit accidentally. It is the procedure behind Who is the Maldivian consumer?, and it is designed to be applied to any consumer category, and eventually to any market with the same characteristics: a small population, a large non-resident or non-household workforce, no recent income survey and no transaction data.

It is deliberately not a model with fitted coefficients. The Maldivian data will not support one. It is a cascade of explicit deductions, each of which can be checked, argued with, or replaced when better evidence appears.

The cascade

POPULATION
    ↓  who is actually here
COMPOSITION
    ↓  nationality, age, location, living arrangement
MEASURED POPULATION
    ↓  who any survey actually observes
LABOUR MARKET
    ↓  who earns, in what sector, under what wage floor
INCOME
    ↓  level, and — far more important — distribution
HOUSEHOLD STRUCTURE
    ↓  the unit that actually buys
LEAKAGE
    ↓  remittances, savings, imports, external spending
DOMESTIC SPENDING
    ↓  what stays and is spent here
SECTOR DEMAND
       exposure by category, not by average

Each arrow is a deduction that must be evidenced separately. The framework’s discipline is that no arrow may be crossed with an assumed constant unless the assumption is stated, tiered and shown in the output.

The five rules

1. The demand unit is not always a person. Housing, utilities, connections, appliances, insurance and broadband are bought by households. Food, telecoms and transport are bought closer to per head. In the Maldives these two denominators have diverged sharply: households grew about 4.1% a year over 2014–2022 against 1.65% for resident Maldivians [OFFICIAL counts; rates OWN-CALC]. Choosing the wrong denominator is the most common and most expensive error in the market.

2. Distribution dominates the mean. Where the top quintile makes 39% of national household spending — and 54% within Malé [OFFICIAL — HIES 2019] — a mean-based market size is not merely imprecise, it is describing a household that does not exist. Every category must be sized against the part of the distribution that actually buys it.

3. Measured is not the same as present. Before any spending arithmetic, the framework asks who the source survey can see. In the Maldives, HIES excludes collective living quarters of ten or more residents, resorts and industrial islands — which is where 79% of enumerated foreign residents live [OFFICIAL]. The measured population and the resident population differ by roughly a quarter, and no adjustment is applied silently.

4. Leakage is a first-class term, not a rounding error. Income earned domestically is not domestic demand. Remittances, imported consumption and external spending all reduce it. Where the leakage cannot be sized — as with Maldivian remittances, which two official series bracket at a fourfold range — the framework reports the range and refuses to pick a midpoint.

5. Vintage is part of the number. An estimate combining a 2019 unit value with a 2026 population is a volume at 2019 prices, and must be labelled as such. Mixing vintages silently is how false precision enters a model.

Evidence classes

Every figure in a Nyra demand estimate carries one of four labels, in addition to its source tier:

The source tiers are the house standard used across all Nyra research — OFFICIAL, TRANSACTION, ASKING, LISTING, INDUSTRY-EST, MEDIA, COMMENTARY, OWN-CALC. A figure may be an official fact, an own-calculation estimate, or an own-calculation scenario; the two labels answer different questions and both are reported.

Sector exposure

The final step converts the cascade into a per-category view. For each sector the framework asks six questions:

1. Population exposure — does demand track head count, household count, or neither?
2. Income exposure — which part of the distribution actually buys it?
3. Migrant exposure — how much of the demand comes from residents no survey measures?
4. Tourism exposure — how much comes from non-residents entirely?
5. Price sensitivity — what does the Engel gradient in the expenditure data imply?
6. Geographic concentration — Greater Malé, atolls, or resort and industrial islands?

A category whose answers are “head count, all quintiles, high migrant, low tourism, low elasticity, dispersed” — staple food — behaves nothing like one whose answers are “household count, top quintile, negligible migrant, low tourism, high elasticity, Malé” — restaurants. Sizing both from the same population figure guarantees that at least one is wrong.

Where this framework stops working

Stating the failure modes is part of the method, not a caveat appended to it.

• It cannot estimate unmeasured consumption. If a population is outside every survey, the framework quantifies how large that population is and stops. It does not impute their spending from wage assumptions.
• It cannot produce income scenarios without a wage series. The Maldives has no earnings index and no income survey after 2019. Any 2040 income path would be an invention compounded for fourteen years.
• It does not forecast. Population projections are the statistical office’s; migration and income paths are labelled scenarios. Nothing here carries a prediction interval, because nothing here has been backtested.
• It degrades with survey age. Every year that passes without a new HIES widens the uncertainty on every expenditure figure it produces, and the framework has no way to correct for that beyond saying so.

What would improve it

In rough order of how much each would change the answers:

1. A new Household Income and Expenditure Survey. Everything on the income and spending side rests on 2019.
2. A survey covering collective living quarters. This is the single largest blind spot in Maldivian economic statistics, and it is a design choice rather than a technical obstacle.
3. A reconciliation of the two remittance series. The gap between them is the largest unresolved parameter in the model.
4. Any wage or earnings series at all. Its absence is why no defensible income scenario exists.
5. National labour-force results. The quarterly survey covers Malé City only; national unemployment has not been measured since the 2022 census.

Where a commissioned engagement can obtain evidence the public record does not contain — employer payroll data, transaction records, point-of-sale panels, operator data — the framework is designed to take those inputs in the same cascade, with the same labelling, and produce materially narrower ranges. That is the difference between what we can publish and what we can find out.

---

Sources

• NBS/MBS — HIES 2019 Income and Expenditure reports
• MBS — Census 2022 (population, households, employment)
• MBS — Maldives Population Projection 2022–2062
• World Bank WDI / MMA Payments Bulletin — remittance series