Tourism flows to Caribbean islands: an empirical note

Abstract

This article examines the determinants of tourism flows to 31 small-island Caribbean economies, from 12 source countries, using an empirical approach that accounts for the inherent heterogeneity of the tourism product and tourism consumers. The results show that tourism flows from the source markets to the Caribbean are income elastic, with the exception of flows from Spain and the USA, but are not very sensitive to price changes.

Keywords:

• tourism demand
• modelling
• relative prices
• Caribbean

JEL Classification:

• N16
• C33
• C31

Notes

¹ The host countries are: Anguilla, Antigua and Barbuda, Aruba, The Bahamas, Barbados, Belize, Bermuda, Bonaire, British Virgin Islands, Cancun (Mexico), Cayman Islands, Curacao, Dominica, Dominican Republic, Grenada, Guadeloupe, Guyana, Haiti, Jamaica, Martinique, Montserrat, Puerto Rico, St. Kitts and Nevis, St. Lucia, St. Maarten, St. Vincent and Grenadines, Saba, St. Eustatius, Suriname, Trinidad and Tobago, and US Virgin Islands. The source countries are Belgium, Canada, France, Germany, Japan, Italy, Netherlands, Spain, Sweden, Switzerland, UK, and the US.

² This is based on Durbarry (2001) and Zhang and Jensen (2005).

³ For example, tourism arrivals and real GDP in the Eastern Caribbean Currency Union (ECCU) are cointegrated and exhibit the following long-run relationship: ln(GDP) = 3.73 + 0.67ln(tourism) + 0.01t, where tourism is the total number of tourists that arrived in the ECCU region, real GDP is the total ECCU real GDP, and t is a linear time trend. The analysis uses annual ECCU data covering the period 1979–2005. In the short run, Δln(GDP) = –0.44(ln(GDP _t–1) – 0.67ln(tourism _t–1) – 0.01t – 3.73) + 0.44(Δln(GDP _t–1) + 0.50Δln(GDP _t–2) – 0.19Δln(tourism _t–3). Panel analysis – based on Pedroni (2004) – involving the six ECCU countries for the period 1996–2005 also support these findings.

⁴ Tourism arrivals data were used due to lack of reliable comprehensive expenditure data.

⁵ Above-trend arrivals are extracted using the Hodrick Prescott (1997) filter as a proxy for the tourism industry's potential. When arrivals are above trend (i.e. a positive gap) we assume that the industry has performed well, and the converse when they are below trend.

⁶ Due to limited long-run data, the short-and-long run effects could not be decomposed.

⁷ Due to data constraints, only a one-year lag is considered.

⁸ The airlift variable used in the estimation only captures scheduled flights.