Festival Economics Dataset

Festival discussion is unusually vulnerable to survivorship bias. Large durable brands remain highly visible, while canceled, dormant or short-lived events disappear from public calendars and eventually from datasets assembled from current websites. Pollstar reported that roughly 90 festivals were called off in 2024, amid discussion of production costs, consumer wallets, weather and changing market conditions.¹ A serious festival dataset therefore needs to retain failure, hiatus and ownership change as first-class observations rather than studying only the events that remain operational.

The Festival Economics Dataset is intended as a longitudinal panel in which one row represents one festival edition. A recurring festival would therefore generate a separate observation for each year, linked to a stable festival entity. This structure allows Live Index to measure changes in duration, location, capacity, ticket price, lineup scale, promoter, ownership and reported attendance over time while preserving the identity of the underlying event.

Core variables

Each edition should record festival name, year, dates, market, site, number of days, number of stages where verifiable, ticket products, all-in price where observable, capacity, reported attendance, attendance definition, sellout status, weather disruptions, cancellations, ownership, promoter, ticketing provider and status in the following year. Artist count and headliner tier can be stored as supplemental programming variables, but automated lineup analysis should preserve source documentation because festival billing and artist cancellations change.

Financial variables require particular caution. Gross ticket revenue should only be reported when derived from reliable box-office records or explicitly labeled estimates. Economic-impact figures should store output, value added, labor income and tax effects separately, along with the model and geography. The Free Press Summer Fest case demonstrates why: the University of Houston study's approximately $14 million output estimate is not equivalent to the festival's ticket gross.²

Attendance definitions

Attendance should never be a single unlabeled field. Festivals may report unique attendees, total admissions across multiple days, scans, tickets sold or modeled attendance. The dataset should store both the reported value and the definition, and observations with unknown definitions should be flagged. The 2012 FPSF record provides a useful historical example: contemporaneous Chronicle reporting cited an estimated 90,000 attendees, while the University of Houston impact model used 81,000.²³ Both can be retained without pretending they are directly comparable.

Failure and dormancy

A festival should receive an annual status even when no edition occurs. Possible states include held, canceled before on-sale, canceled after on-sale, postponed, dormant, permanently ended, renamed or merged. This enables survival analysis and allows Live Index to test whether festival attrition is concentrated by scale, ownership, region or price tier.

Research use

The dataset should support studies of ticket inflation, festival survival, market saturation, weather exposure, ownership consolidation, visitor draw and the relationship between site type and longevity. It can also connect to the Live Music Economic Impact Index where credible impact studies exist. The objective is not to produce a universal “best festival” score but to create a research base that makes claims about festival economics reproducible.