Events and Meetings of Italian Statistical Society, Advances in Latent Variables - Methods, Models and Applications

Font Size: 
A multilevel approach for repeated cross-sectional data: an application on auction prices.
Lucia Modugno, Silvia Cagnone, Simone Giannerini

Last modified: 2013-06-14


Repeated cross-sectional data consist of observations on individual survey respondents drawn from the same context at many different time-points. This paper proposes a multilevel approach for this peculiar data structure, by treating individuals as clustered within time-points. The key difference of the proposed multilevel framework with the traditional one is the possibility of modelling the time effects by allowing for serial correlation among level-2 units. We derive and implement full maximum likelihood estimators with known desirable properties, and we obtain model estimates through an EM iterative algorithm. The work has been motivated by the analysis of the first world database of Tribal artworks sold in the most important auction worldwide. We apply the proposed model on auction prices. The results show that this specification improves considerably over existing proposals both in terms of fit and prediction.

Full Text: PDF