Income moves Maldives demand; price stays hidden in the data

Executive summary

The descriptive reports show that the Maldives’ source markets move. This one asks why — and how much. For each major market we fit an ARDL error-correction model of arrivals on source-country income (real GDP per capita) and a relative price (the cost of the Maldives versus home, in the traveller’s own currency), on the clean pre-COVID years, and run a Pesaran bounds test for a stable long-run relationship.

Source markets — arrivals, 12-month rolling total

Maldives · monthly · rolling 12m sum

ChinaIndiaRussiaUKGermanyItaly

Data table

Arrivals by source market, 12-month rolling totals

Period	China	India	Russia	UK	Germany	Italy
Jan 1999	—	118,179	—	378,042	729,768	608,251
Jan 2000	—	120,459	—	439,076	766,555	686,373
Jan 2001	—	123,269	—	500,372	796,656	771,168
Jan 2002	—	127,223	—	558,065	818,140	846,315
Jan 2003	106,654	130,677	—	637,284	843,480	948,180
Jan 2004	108,728	133,871	67,220	735,775	864,306	1,030,075
Jan 2005	109,852	133,241	81,653	804,756	859,395	1,053,961
Jan 2006	124,074	133,689	102,584	886,361	863,245	1,127,016
Jan 2007	152,705	138,363	131,729	973,853	863,188	1,192,157
Jan 2008	191,083	144,787	175,182	1,045,386	857,740	1,242,048
Jan 2009	250,341	148,531	210,292	1,099,784	850,752	1,275,897
Jan 2010	369,874	162,553	255,237	1,158,213	851,543	1,286,217
Jan 2011	569,861	181,910	315,745	1,198,802	855,563	1,280,848
Jan 2012	798,462	203,015	378,515	1,219,138	876,272	1,237,179
Jan 2013	1,129,361	232,518	451,011	1,227,856	903,721	1,179,293
Jan 2014	1,487,945	266,728	509,769	1,236,183	938,837	1,122,200
Jan 2015	1,841,615	307,594	541,984	1,234,969	973,207	1,047,512
Jan 2016	2,154,847	363,550	570,431	1,222,821	1,006,621	987,670
Jan 2017	2,460,094	436,309	617,780	1,239,534	1,062,948	1,006,422
Jan 2018	2,729,253	514,712	666,760	1,246,141	1,109,650	992,790
Jan 2019	2,990,260	663,415	718,284	1,247,182	1,168,942	1,011,887
Jan 2020	2,983,389	709,712	730,693	1,183,081	1,136,137	954,754
Jan 2021	2,923,172	985,649	913,101	1,139,908	1,162,410	892,771
Jan 2022	2,815,403	1,201,275	1,065,944	1,205,059	1,218,734	900,777
Jan 2023	2,798,568	1,379,490	1,211,154	1,256,545	1,263,308	935,974
Jan 2024	2,827,927	1,478,574	1,369,980	1,346,414	1,322,203	1,018,864
Jan 2025	2,818,796	1,572,185	1,572,262	1,462,717	1,396,427	1,114,424

The result is honest and split down the middle. A genuine long-run demand relationship exists in four of the six markets — China, Russia, the UK and Germany — and every one of them is strongly income-elastic. But the price side refuses to identify on this data: the signs flip market to market, and only two of six come out with the textbook negative. Income is what free annual data can see; price is not.

What the model finds

Market	Long-run relationship	Income elasticity	Price elasticity	Adjustment / yr
China	yes (5%)	+3.1	+2.6 ⚠	30%
Russia	yes (5%)	+3.9	−0.3	64%
United Kingdom	yes (5%)	+3.9	−0.1	41%
Germany	yes (5%)	+2.3	+0.6 ⚠	56%
India	none found	—	—	—
Italy	none found	—	—	—

⚠ = a positive price elasticity, i.e. the wrong sign for a price effect — a signal the model cannot trust here.

Income — robust and large. Every identified market has an income elasticity well above 1: arrivals are a luxury good, and they grew faster than income as source-country middle classes gained not just the money but the access to travel long-haul. The magnitudes (roughly +2 to +4) run higher than the tourism literature’s usual +1.5 to +2 — partly that access effect, partly the short samples. Read them as “high and positive,” not as precise point estimates.

Adjustment — sensible. When arrivals drift from their income-implied level, they close 30–64% of the gap each year — equilibrium restored in roughly one-and-a-half to three years. That is exactly how a demand relationship should behave.

Price — not identified. Only the UK and Russia produce a correctly-signed (negative) price elasticity, and both are small and imprecise; China and Germany come out positive. A free, annual relative-price proxy built from national CPI and exchange rates is simply too crude to isolate price sensitivity — and there is no free monthly real-exchange-rate series to do better.

The honest limit

This is the structural layer the platform was built toward, and it is worth being exact about what it delivers. It can establish that Maldives demand is real, cointegrated and strongly income-driven in most major markets. It cannot, on free data, deliver a trustworthy price elasticity, and it finds no stable relationship at all for India (whose series is dominated by its post-2010 boom and 2024 reversal) or Italy.

The obvious fix — more data, at quarterly frequency with a sharper real-exchange-rate price term — was tried and blocked by the same wall: free quarterly GDP exists for the UK, Germany, Italy and India, but not for China or Russia, the two emerging markets where price and income detail matter most. Interpolating their annual income to quarterly would manufacture the very variation the model would then claim to measure — so we don’t. Pinning down price, and tightening the income estimates, would take what stays out of reach: proprietary forward-booking and rate data, real quarterly macro for the emerging markets, and a longer clean run without a pandemic in the middle.

Methodology

Per market, an ARDL(1,1,1) unrestricted error-correction model of log arrivals on log real GDP per capita (World Bank NY.GDP.PCAP.KD) and a log relative price (Maldives CPI × exchange rate ÷ source CPI; World Bank FP.CPI.TOTL, PA.NUS.FCRF), estimated on the contiguous pre-COVID window (~1988–2019, n=25–32). Cointegration by the Pesaran (2001) bounds test, case III; long-run elasticities and the error-correction speed read from the levels terms. Reproducible via scripts/demand_elasticities.py.

Source: Avé Intelligence · MMA arrivals by country · World Bank WDI · n=25–32 annual obs per market, ~1988–2019.

---

Sources

• Avé Intelligence
• Maldives Monetary Authority (arrivals by country)
• World Bank WDI (GDP, CPI, exchange rates)