Valuing cultural world heritage sites: an application of the travel cost method to Mont-Saint-Michel

ABSTRACT

This article, using the zonal version of the travel cost method (ZTCM), provides the first economic valuation of recreational trips to Mont-Saint-Michel (MSM). The MSM was designated as United Nations Educational, Scientific and Cultural Organization World Heritage Site in 1979, and is the most visited coastal site in France. The potential effects on consumer surplus (CS) estimates of some aspects of the ZTCM are considered, namely the treatment of the multiple purpose/destination (MP/D) trips, the inclusion of the opportunity cost of time (OCT) in the recreation demand model and the heterogeneity in the preferences. The heterogeneity in the preferences is examined through a random parameters Poisson model (RPPM). The RPPM generalizes the standard Poisson model (SPM) by allowing coefficients of explanatory variables to vary randomly across geographical units rather than being fixed. Our results show that (1) substantial CS values are associated with the MSM; (2) excluding/ignoring MP/D trips result in biased CS estimates; (3) omitting OCT does not substantially bias CS estimates; (4) the RPPM provides richer information about the impact of explanatory variables on the demand for trips than the SPM, but (5) leads to statistically similar CS estimates than the SPM.

KEYWORDS:

• Recreation demand
• nonmarket benefits
• zonal travel cost method
• count models
• Mont-Saint-Michel

JEL CLASSIFICATION:

• D12
• Q26

Acknowledgements

This survey was carried out within a research project funded by the French Ministry of Ecology, Sustainable Development and Energy (Liteau programme). Thanks are due to the following partners: IUEM (Université de Bretagne Occidentale) for funding an internship, LETG-GEOMER (Université de Bretagne Occidentale) for providing data from the Bountîle Observatory, and SMBSM (Syndicat Mixte de la Baie du Mont-Saint-Michel) for providing practical assistance to the implementation of the field surveys. Special thanks to Marie Guingot for data entry and cleaning. We are grateful to Stephen Hynes, Ken Baerenklau, William Greene and the reviewers for helpful discussions and comments. The authors are sole responsible for the content and views expressed in this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The exclusion of individuals reaching the Mont-Saint-Michel with the help of a collective motor vehicle avoided us to make some rather strong assumptions about their travel costs that are found in the literature like ‘the air fare, rail fare or bus fare is for the economy class booked four weeks in advance’.

² The cost per kilometre is published annually by the fiscal administration. In 2011, it was 0.52€ per kilometre. It includes four factors: the vehicle depreciation, maintenance, fuel and insurance costs. In this article, the insurance cost was excluded from the travel cost calculation as it is generally fixed by period of time and do not vary on the margin (the marginal kilometre). It was estimated by simply dividing €0.52 by 4, resulting in €0.13 per kilometre. Parameter c was thus computed by subtracting €0.13 from €0.52. We also estimated the models using the full cost per kilometre (€0.52). These results are available from authors upon request.

³ This choice was also based on feedback from experts arguing that they have no information about which sites close enough that would be appropriate substitutes for the MSM.

⁴ We applied an overdispersion test based on simple regression techniques, proposed by Cameron and Trivedi (1990) and shown in Greene (2007), within the context of the Poisson model. The results suggested that overdispersion is not present in the data (see Tables 2 and 3). This suggestion was then confirmed by the results of NB models, which showed that the overdispersion parameter was not significant for either SP/D visitors or all domestic visitors (these results are available upon request). In addition, Poisson models with robust SEs provided similar results than NB models in terms of goodness-of-fit measures (McFadden pseudo R²) but better ones in terms of CS. Such results are not specific to our study (e.g. Von Haefen and Phaneuf 2003; Baerenklau 2010; Willis et al. 2012). This led Von Haefen and Phaneuf (2003) to suggest that the Poisson model is more attractive for policy purposes relative to the NB model.

⁵ Bhat (2003) demonstrated that this number of draws produce accurate parameter estimates.

⁶ Similar to Whitehead et al. (2016), the parameter for the travel cost variable was first specified as random assuming different distribution functional forms (i.e. normal, uniform and triangular distributions), but the SD parameter was not statistically significant. So we estimate it as fixed parameter.

⁷ Due to the exceptional character of the restore the maritime character of MSM project, its cost does not meet the definition of preservation and maintenance costs under the BECO project. For this reason, it was excluded from the calculation.

⁸ We thank one reviewer for this comment.

⁹ The latent class and random parameters models are theoretically equivalent (Baerenklau 2010).

Additional information

Funding

This work was supported by the French Ministry of Ecology, Sustainable Development and Energy (LITEAU program).