Auction Catalogue Narratives, Moral-Historical Framing, and Auction Outcomes for Adrian Ghenie Lots

This repository contains the data and code for a mixed-method study of how leading auction houses narratively frame morally burdened historical references in the sale of Adrian Ghenie’s paintings and how such framing relates to auction outcomes. The dataset links publicly available, author-archived lot texts (catalogue essays and associated online lot-page descriptions) from Christie’s and Sotheby’s to structured auction variables (sale date, estimates, realized price where available, dimensions, and work metadata). It also includes transparent dictionary-based measures of moral-historical language (e.g., moral_flag, moral_token_count, moral_terms_hit, and an ordinal moral_intensity score), plus scripts used to generate the analytic variables and reproduce the reported models and diagnostics.
Only publicly accessible catalogue materials were used; no private client communications or non-public auction-house documents are included. The corpus was collected and archived during a defined access window (November 2025). The analytic sample used in the paper contains 106 lots after standard exclusions for missing or nonpositive estimates/prices and missing key controls. These materials support reproducibility and enable reuse for research on valuation, narrative framing, moralization in markets, and cultural intermediaries. Lot texts were manually collected from publicly accessible online catalogues and archived by the author(s). Moral-historical framing was operationalized using a transparent dictionary approach: term hits were recorded per lot (moral_terms_hit), counted (moral_token_count), flagged (moral_flag), and summarized using an ordinal intensity rule (moral_intensity, 0–3) based on the most severe term-layer present in each text. Auction outcome variables were compiled at the lot level and merged with the archived text and moral measures. Analysis scripts reproduce data cleaning, variable construction, and regression models reported in the associated paper.