The paper presents alternative multinomial specifications for analysing discrete choice preferences expressed over different supply options of a multi-dimensional and multi-attribute cultural heritage site. The database derives from a stated preference choice modelling experiment carried out on visitors of the Galleria Borghese Museum in Rome, Italy, a worldwide known cultural heritage site. Econometric analysis relies both on Conditional Logit and on mixed logit models.
First, alternative Conditional Logit specifications are investigated, testing relative performances and consistency. An analysis based on sample segmentation by socio-economic elements is also presented, as an alternative option of addressing heterogeneity. Both approaches shed light on which socio-economic factors determine choice heterogeneity in the sample, affecting the parameters of the model. A systematic test of IIA violation is carried out using Hausman statistics. Then, models not based on the IIA assumption are introduced. Heterogeneity is addressed both by classic and systematic approaches.
The main result arising is that hybrid forms of conditional logit, which incorporate heterogeneity by the classic approach of adding interaction socio-economic terms, do not violate the IIA assumption and show outcomes which are consistent with a priori expectation based on microeconomic theory. Thus, specifications relaxing IIA do not outperform the Conditional Logit model. Most attributes defined arise as significantly affecting choices: choice experiments capture both the relative relevancy of attributes for users, and the latent heterogeneity.
Finally, marginal willingness to pay point estimates and economic surplus figures are presented, showing that changes in conservation and additional services, multi-media services and temporary exhibition, with respect to the status quo, are relevant attributes for visitors, who reveal positive Willingness to pay (WTP) at the margin.
Choice experiments confirm to be a practical and effective tool for non-market valuation based on stated preference, aiming at informing and justifying - on economic grounds- demand led policies in the cultural sector. It is specifically targeted to evaluation of mixed good with relevant private components. Information on marginal WTP for changes in supply, and on factors determining visitor's preference heterogeneity, may be extensively studied.
Nonetheless, given the complexity in design and estimation, each case study needs a comprehensive econometric analysis, which makes use of many different specifications and a wide array of test statistics.
Further research is encouraged in order to achieve generality of results, for the setting of choice experiment as an effective tool for cultural policy making.