Neither a Picasso Nor a da Vinci: Pricing Artwork of Emerging Artists
(A2024-118480)

AUTHORS

Sharmistha Sikdar, Tuck School of Business at Dartmouth; Ishita Chakraborty, UW Madison; Nika Dogonadze, Takeup.AI

ABSTRACT

Emerging artists with an uncertain reputation who sell artwork for aesthetic purposes often find it difficult to determine the right price. Online advisory sites advise pricing using heuristics like price per square inch and artist’s labor rate. These sites do not consider visual cues, painting descriptions, artist’s past sales successes, etc. that might affect a buyer’s willingness to pay (WTP). We use scraped data from Etsy.com on prices, images, and other characteristics of sold artwork and artists to build a scalable, multi-modal neural network (NN) model to predict customer WTP. Our model uses multiple modalities, i.e., structured inputs like artwork dimensions, and unstructured textual data from painting descriptions and visual images of the artwork. We perform a WTP experiment and find that our model predicts WTP better than the online advisory sites. Our multi-modal NN model can help novice artists determine prices for their artwork and be more successful on online platforms.