The short version

National statistics describe groups, not shops. Two official sources look useful to a Fiji supermarket: the 2019-20 household income and expenditure survey (HIES), and the Bureau of Statistics' wholesale and retail trade series for 2024 and 2025. Neither can say what is happening inside a particular store. The survey describes how groups of households spend, region by region. The trade series adds up registered businesses across wholesale, retail and motor vehicles. Both were also revised after first publication, which is a finding in itself.

This is a methodological investigation. It contains no store data and reports no sales results, advertising effects or causal claims.

1. A correction, not a reversal

Household survey · Fiji · 2019–20

Living below the poverty line

  • Original · August 2021
  • Revised · April 2022
  • National Original 29.9%Revised 24.1%
  • Urban Original 20.4%Revised 14.0%
  • Rural Original 41.5%Revised 36.5%
The April 2022 coding correction recalculated the same survey, using the same households and fieldwork. All three rates fell. Rural poverty stayed above urban poverty, and the order of the four divisions stayed the same.

In August 2021 the Fiji Bureau of Statistics (FBoS) reported that 29.9% of Fijians lived below the poverty line in 2019-20. In April 2022, after a cross-check by the World Bank, the figure was restated as 24.1%. According to the two published notes, the cause was a coding fault: for four non-food categories (utilities, communications, domestic services and education), a household counted as spending only if it had spent on every sub-item. That affected nearly nine in ten sampled households. Food was unaffected.

The two published measurements describe the same survey, the same households and the same fieldwork. The second restates the first. It is not a new observation.

Poverty rates in the original and revised publications
Area Original (August 2021) Revised (April 2022)
National 29.9% 24.1%
Urban 20.4% 14.0%
Rural 41.5% 36.5%

Every level fell, and nothing reversed. Rural poverty is higher than urban in both versions, the order of the four divisions is unchanged, and the poverty line itself was almost unchanged. The gap between the rural and urban rates is slightly wider after the correction: 22.5 points against 21.1 (our arithmetic on the published rates). Estimated average nominal consumption per adult equivalent was revised up by about 8.5%, by about 10% in urban areas and about 6% in rural areas. Nominal food expenditure was unchanged by the coding correction.

The lesson is about reading official numbers. The pattern survived the correction, but the level was overstated by 5.8 percentage points for roughly eight months, in a survey that was carefully run and then cross-checked. When a national figure is used to size a market, the level deserves more caution than the pattern.

2. What a household survey can and cannot see

The HIES visited about 6,000 households in 600 areas between February 2019 and February 2020. The households fall into seven groups by division and by urban or rural. Results are designed for those groups. The report states that provincial and tikina estimates are not supported, and it measures consumption excluding durable goods such as vehicles and appliances.

So the survey can support a sentence like this one: households in the Eastern division spent a far larger share of their budget on food (about 54%) than households in the Central division (about 39%). It cannot support a sentence like this one: shoppers at one supermarket are poorer than shoppers at another. The survey has no town, catchment or store in it. In the tables we read, spending is not broken down by shop.

Published averages also drift in the retelling. On 14 September 2021 the Fiji Times reported that 29 per cent of household income in rural areas was spent on vegetables. The report says vegetables were 29.4% of rural food spending. A share of a share became a share of income. We cite this one example and do not claim it is typical.

3. What an average hides

An average describes a group, not a member of it. The report gives two summaries for each group: the mean, which is total spending divided by the number of households, and the median, which is the spending of the household in the middle when all are ranked. Nationally, household consumption has a mean of FJ$11,961 a year in real terms and a median of FJ$10,768. The urban mean is FJ$13,521 and the rural mean is FJ$9,941.

Each is one figure standing in for thousands of households. Neither shows how spending is spread across them, and neither identifies which households sit above or below it. Two groups with the same mean can look very different inside.

So the urban mean is accurate for urban households as a group, and it tells a retailer nothing about any one urban store. A store in a high-spending suburb and a store in a low-spending one can both sit inside the same urban figure. The survey reports its figures per household and per adult equivalent, which adjusts for household size and age. We use the per-household figures here and do not set the two measures against each other.

4. A national number about businesses is not a number about shoppers

The second source counts businesses, not households. Each quarter FBoS surveys registered wholesale, retail and motor-vehicle businesses (about 80% of sector turnover, according to its releases) and publishes total sales. For 2025 the total is FJ$7,147.3 million, up FJ$376.0 million or 5.6% on 2024.

That figure measures aggregate trade turnover across the sector. It is not Fiji's grocery market, and it is not what consumers spend on groceries. The total combines wholesalers, vehicle dealers and retailers, so the same goods may be counted at more than one stage of the supply chain. It includes fuel and vehicles.

The closest published line to supermarket trade is "non-specialised stores with food, beverages or tobacco" (class 47111): FJ$865.8 million in 2025, about 12% of the total. The class describes stores selling a general range of goods with food, beverages or tobacco predominating. It is not a list of supermarkets. It may include other store types, it may omit food retailers classed elsewhere, and the survey covers registered businesses only. We have not confirmed from the releases which stores FBoS places in it. It is a category total: a single store can grow while the category falls, or the reverse, and the total would not show it.

5. The same year, two annual figures

Wholesale and retail trade · Fiji

One year, two published totals

Annual trade turnover · FJ$ million

  • 2024 originalAnnual 2024 · 24 Mar 2025 6,890.2
  • 2024 restatedAnnual 2025 · Mar 2026 6,771.3
  • 2025 new periodAnnual 2025 · Mar 2026 7,147.3

−FJ$118.9m

The entire reduction was in the December 2024 quarter, from FJ$1,876.7m to FJ$1,757.8m. The other three quarters match across the two annual publications.

Quarterly publications also report 2024 figures. The March 2025 quarterly release carried the lower December 2024 figure with a source-revision note. An independent factual review identified the June 2025 revision marker in the September 2025 quarterly release. The reasons for the revisions are still unknown.

Two annual FBoS publications give different totals for 2024. The 2024 annual release (24 March 2025) gave FJ$6,890.2 million. The 2025 annual release (March 2026) gives FJ$6,771.3 million for the same year, FJ$118.9 million lower. Quarterly releases published 2024 figures as well, so this comparison covers two of the publications, not every one.

The whole difference sits in one quarter. March, June and September 2024 are identical in the two annual releases. December 2024 was first published as FJ$1,876.7 million and now stands at FJ$1,757.8 million. The March 2025 quarterly release, dated 23 June 2025, already carried the lower figure, marked as revised with the note "revised due to revision from source". We do not know what the source was or what changed, and we have not guessed.

Two things follow. First, the 5.6% growth figure for 2025 is valid because it compares 2025 with the restated 2024 inside one release. Subtracting the original 2024 figure would give 3.7% and would mix two releases. Second, later quarters moved too. The September 2025 quarterly release already marked the June 2025 figure as revised, with a source-revision note. The 2025 annual release then shows June at 1,734.9 against 1,749.5 when first published, and September at 1,848.7 against 1,869.5, also marked revised. Any recent quarter should be treated as provisional.

This is the poverty lesson again in a different place. A revised figure restates something already measured. A new figure measures a new period. Mixing them can produce a trend that does not exist.

6. Percentages and dollars together

A growth rate and a dollar change answer different questions, and both are needed. The rate shows how fast a line grew relative to its own size. The dollar change shows how much it added to the total. A small line can grow quickly and add little. A large line can grow slowly and add a lot.

Inside the 2025 increase of FJ$376.0 million, books, newspapers and stationery rose 28.8%. It added FJ$15.3 million, about 4% of the increase. Motor vehicle sales grew 25.5% and added FJ$89.9 million, about 24%.

Selected contributions to the 2025 increase
Line named in the 2025 release Change, FJ$ million Share of the 2025 increase
Motor vehicle sales +89.9 23.9%
Wholesale food and non-alcoholic beverages +58.5 15.6%
Hardware +43.4 11.5%
Wholesale fuels +42.5 11.3%
Retail automotive fuel +21.5 5.7%
Non-specialised food stores (47111) +19.6 5.2%
Books, newspapers and stationery +15.3 4.1%

The supermarket-type line was about 12% of the 2025 total and supplied about 5% of its growth. These are the lines the release names, not a full breakdown, and because wholesale and retail lines can count the same goods at different stages, the shares are not consumer spending. The point is narrow. Read the rate and the dollar change together before saying where growth came from.

7. Why neither number reaches the shelf

From national context to store evidence

What changed at this store?

  • Household surveyHouseholds grouped by region
  • Trade seriesRegistered businesses grouped by trade
One store, one week Neither source can tell us what happened at its tills or why.

Evidence a retailer would need

  • Till recordsBy store and week
  • Prices and stockTo account for changes and empty shelves
  • Customers or catchmentDefined locally, with consent for customer data
  • Designed comparisonSimilar stores or periods chosen in advance
The study holds none of these store-level records. A designed comparison can estimate a possible effect, with limits on what it establishes about cause.

Put the two sources together and their limits are plain.

  • Different units. The survey counts households, grouped by region. The trade series counts businesses, grouped by line of trade. Neither counts a store's customers or a store's tills.
  • Different places. The survey cannot see below its seven groups. A supermarket's customers come from a town or a few suburbs.
  • Different times. The survey covers February 2019 to February 2020. The trade figures cover 2024 and 2025. They cannot be laid side by side as one moment.
  • Averages and totals. An average household hides the spread across households. A category total hides the spread across stores. Any one store can sit well above or below either.
  • Revisions. Both sources were corrected after first publication, one for a coding fault and the other for a reason the releases we read do not specify.

Regional differences in household spending, and growth in a trade category, are therefore context. They are not evidence of how any particular supermarket is performing.

A retailer who wants that answer needs evidence from inside the business:

  • till records by store and week, ideally by product and promotion;
  • price and stock records, so a sales change is not confused with an empty shelf;
  • a defined customer base or catchment, for example loyalty data held with consent;
  • a comparison designed in advance, using similar stores or periods that did not receive the change, so that any difference can be examined as a possible effect.

Each is the retailer's own data or a designed study, and no national figure replaces it. Even a well-designed comparison points to a possible effect and its likely size. It does not prove cause on its own, because stores and periods differ in ways that also move sales, and its limits should be stated with the result. We hold none of this and report no store results.

What we could not settle. The reason for the December 2024 restatement. One measurement question inside the HIES, recorded in the evidence notes. Whether the next HIES, with fieldwork planned for March 2025 to March 2026 and no results found, changes the picture. We will revisit the study if any of these resolves.