Natural amenities and tourism employment: A spatial analysis

Abstract

This thesis examines the spatial linkages between natural amenities and tourism employment when spatial spillover effects are taken into account under three neighbourhood structures. To the author’s knowledge, tourism employment and natural amenity spillovers have not previously been examined in a non-geographic spatial context.To address this research gap and explore the non-geographic spatial spillovers in tourism employment and natural amenities, spatial models are developed with different neighbourhood structures. In addition, the thesis examines: tourism clusters; the measurement of cluster proximity; and characteristics of tourism employment within and across clusters.This thesis developed an empirical model at the local government level for Queensland, Australia. The model incorporates 13 natural amenity variables and four other variables that explain regional tourism employment. The analysis is conducted for 74 local governments in Queensland using cross-sectional data compiled for 2011.Three specifications of a Spatial Durbin Model (SDM) are implemented. The specifications use three alternative weight matrices to reflect three different proximity dimensions or neighbourhood structures. A non-spatial linear model (NSLM) estimated by least squares is used as the base model. The three weight matrices specifications are proportional to: a) geographic contiguity, to capture geographic spillovers; b) the share of employment in the tourism industry; and c) the proportion of tourism employment to total employment. The last two are measures of economic distance.To identify tourism employment clusters, the study uses a K-mean cluster analysis technique. Factor analysis is used to summarise 17 variables selected from a review of theoretical models of tourism employment. Based on computed factor scores, the main cluster attributes are identified.The results suggest that independent of the specification (spatial and non-spatial), internet penetration, regional population, the number of regional parks and state forests and World Heritage areas are statistically significant. This is taken as evidence of their importance as factors that influence tourism employment in a region. Most importantly, spillovers exist; not only between traditional geographic neighbours, but also between neighbours of economic proximity where neighbours are those that have a similar profile in their share of tourism employment.When the spatial autocorrelation parameter is negative and significant between geographic neighbours, it reflects competition; however, when it is positive and significant between economic neighbours (regions with similar share of tourism employment) it indicates a collaborative effect between the regions with similar tourism performance. The results further suggest non-geographic proximity can explain tourism employment clusters. Characteristics such as “urbanness”, communications capacity, natural attractions, level of agriculture and percentage of indigenous population are shared within clusters.The conclusion is that non-geographical proximity dimensions can be highly informative when developing plans or strategies to maximise spillovers and clustering in regional tourism.