The effects of tourism market diversification on CO₂ emissions: evidence from Australia

ABSTRACT

For the first time in the tourism and environment literature, this study investigates the CO₂ emissions effect of the market diversification of tourist arrivals. Theoretically, tourism market diversification has two opposite potential effects on CO₂ emissions, depending on its scale and composition effect. It may increase CO₂ emissions by expanding the scale of tourism-related activities or decrease CO₂ emissions by expanding the share of developed/less-polluted source countries in the destination. By utilizing the econometrics of time series data, we tested the impact of tourism market diversification on CO₂ emissions in Australia over the period 1976–2019. Our findings show that tourism market diversification has pro-CO₂ emissions effects in the long run for Australia. These findings contribute to our understanding of the environmental impacts of tourism market diversification and help in policy formulations in Australia.

KEYWORDS:

• Tourism market diversification
• CO₂ emissions
• bootstrap ARDL
• Herfindahl–Hirschman index
• Australia

JEL Classification:

• Z30
• Z39
• Q53
• C22

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 We can test the hypothesis using the simple bilateral correlation between TMD and the contribution of tourism industry to GDP and employment. A positive correlation between two variables indicates that the market diversification policy increases the scale of tourism-related activities.

2 To test this hypothesis, we calculate the weighted average CO₂ emissions per capita of the source countries in the destination's tourist basket as follows: $\mathrm{A}\mathrm{C}\mathrm{O}2\mathrm{P}{\mathrm{C}}_{\mathrm{t}}=\sum _{\mathrm{k}=1}^{{\mathrm{H}}_{\mathrm{t}}}{\displaystyle \frac{\mathrm{T}\mathrm{O}{\mathrm{U}}_{\mathrm{k},\mathrm{t}}}{{\sum }_{\mathrm{k}}\mathrm{T}\mathrm{O}{\mathrm{U}}_{\mathrm{k},\mathrm{t}}}\mathrm{C}\mathrm{O}{2}_{\mathrm{k},\mathrm{t}}}$Here, $\mathrm{T}\mathrm{O}{\mathrm{U}}_{k,t}$ is the number of tourist arrivals from source country k to destination; ${\sum }_{\mathrm{k}}\mathrm{T}\mathrm{O}{\mathrm{U}}_{k,t}$ is total tourist arrivals to destination in year t; $H_t$ is the number of source countries of foreign visitors to the destination; and $\mathrm{A}\mathrm{C}\mathrm{O}2\mathrm{P}{\mathrm{C}}_{\mathrm{t}}$ is weighted average CO₂ emissions per capita of source countries in the destination’s tourist basket. The higher values of the measure indicate the share of high-polluted source countries increased in the destination’s tourist basket. A negative correlation between tourism market diversification and $\mathrm{A}\mathrm{C}\mathrm{O}2\mathrm{P}{\mathrm{C}}_{\mathrm{t}}$ measure indicates the diversification increases the share of less-polluted source countries in the destination.

3 Australia welcomed 9.4 million international visitors, resulting in $45 billion international tourism spending (Australian Bureau of Statistics, 2020). Tourism generated 666,000 jobs (5.2% of total employment), contributed $60.8 billion (3.1% of GDP), was the source of 10% of all exports, and engaged 1 in 8 Australian businesses in 2019. However, Australian tourism has been negatively affected by COVID-19 with a decrease of 38% in tourism GDP, 20% in tourism employment, 40% in tourism gross value added, 34% in internal tourism consumption, and 24.6% decrease in tourism labour productivity (Australian Bureau of Statistics, 2021).

4 It should be noted that the index can be calculated using the number of overnight stays, but there is no information about the variable by country of origin.

5 The critical values of the three tests for F_A-statistics, F_B-statistics, and t-student are computed by a bootstrapping procedure developed by McNown et al. (2018).

6 The long-run effects of regressors – i.e., $\ln \left(\mathrm{T}\mathrm{O}\mathrm{U}\mathrm{M}{\mathrm{D}}_t\right),\ln \left(\mathrm{Y}{\mathrm{P}}_t\right),\ln \left(\mathrm{N}\mathrm{R}\mathrm{E}\mathrm{C}{\mathrm{P}}_t\right),$ and $\ln \left(\mathrm{I}\mathrm{N}\mathrm{F}\mathrm{D}\mathrm{I}{\mathrm{F}}_t\right)$ – are calculated as $-{\vartheta }_1/\delta$, $-{\vartheta }_2/\rho$, $-{\vartheta }_3/\delta$, and $-{\vartheta }_4/\delta$, respectively, and their standard errors are computed using the delta method. Here, $\delta$ is the coefficient of the error correction term (ECM(-1)) and measures the adjustment speed of $\ln \left(\mathrm{C}{\mathrm{O}}_2{P}_t\right)$ in the model to move toward an equilibrium level. The first difference terms measure the effects of explanatory variables on $\ln \left(\mathrm{C}{\mathrm{O}}_2{P}_t\right)$ in the short run, and the optimal numbers of lags of the variables are determined using AIC.

7 The results are shown in Table A1 of Appendix.

8 The residual diagnostic tests, including Breusch-Godfrey Serial Correlation LM test, ARCH heteroskedasticity, and Jarque-Bera normality test (P-value of test statistics), confirm no heteroskedasticity, no serial correlation, and normality of OLS estimated residuals. In addition, the Ramsey RESET test, CUSUM, and CUSUM of squares tests confirm the stability of the estimated model. We do not report the results of the test, but they are available upon request.

9 To save the space, we did not report the calculation results of the $\mathrm{A}\mathrm{C}{\mathrm{O}}_2\mathrm{P}{\mathrm{C}}_t$ measure and bilateral correlations between $\mathrm{T}\mathrm{O}\mathrm{U}\mathrm{M}{\mathrm{D}}_t$ index and $\mathrm{A}\mathrm{C}{\mathrm{O}}_2\mathrm{P}{\mathrm{C}}_t$ measure, total number of tourist arrivals, and the contribution of travel and tourism industry to Australia’s GDP and employment. However the results are prepared upon request.