SMU DataArts - Cultural Data Profile


Image courtesy of SMU Meadows School of the Arts. Photo by Kim Leeson Photography. Image courtesy of SMU Meadows School of the Arts. Photo by Kim Leeson Photography.

When Will Arts Attendance Return?

Encouraging Signs from Three "What-If" Scenarios this Holiday Season

Will COVID-19 cases continue to decline or is a new wave coming with winter? Will confidence in health and safety protocols allow performing arts organizations to reopen successfully? Or will ongoing COVID-19 concerns continue to wreak havoc on the road to recovery? There are encouraging signs. At the same time, TRG Arts and Purple Seven recently reported that as of July 2021, holiday productions were tracking well behind 2019–20 in ticket sales and revenues for most theatres.


None of us has a crystal ball to predict the future, but data can offer some insights as to where things could go. In this blog, we first examine the historic impact of COVID on performing arts ticket sales and then we use the data to simulate three plausible “what-if” scenarios – realistic worst-case, realistic best-case, and idealized best-case – to predict the impact of each scenario on ticket sales at mid-sized to large performing arts organizations through the remainder of 2021. The discrepancy between the highest and lowest scenario estimates amounts to an annualized $1.5 million difference in ticket sales for the average performing arts organization in our sample, a sum worthy of consideration. These estimates are intended to provide organizations with guideposts as they prepare their own financial scenarios and navigate emergence from the pandemic, taking into account local COVID conditions.


We begin by building a model that predicts purchase transactions at 51 performing arts organizations across the country, which ranged from $1 million to $167 million in total annual expenses pre-pandemic. They are located in 34 different metropolitan statistical areas and 22 different states in the US. Figure 1 plots the historical values that are normalized on a scale of 0-100 for the characteristics, or “variables,” that vary by month. This way, you can see when each variable was at its maximum (100%) over the period as well as its level of volatility.

For example, ticket purchases were historically lower in the summer months and January. After the arrival of COVID, ticket purchases plummeted to about 5% of their peak level in August and September 2020, whereas restaurant employment trends bottomed out in April 2020 at roughly 50% of their peak level. The trends indicate that state vaccination rates started climbing and COVID case rates declined in February 2021; household ticket purchases, average ticket prices, and the number of performances offered followed suit, as did restaurant employment levels.


Predictive models work by learning patterns that exist in real, historical data, then using those patterns to create algorithms to predict outcomes. You train the model and test for accuracy by comparing predicted results to actual results. Our model is trained on 11.7 million ticket sales that occurred over 42 months from January 2018 through June 30, 2021, with aggregated data provided by our partner TRG Arts. Transactions for each organization were summed up for each household census tract (HHCT) on a monthly basis. We then ran a model predicting the number of transactions each month that incorporates the following influences:

  • organization characteristics (budget size, number of performances, prices)
  • household characteristics associated with the transactions inferred from census data for the household tract (number of households, income, education, age)
  • market characteristics (e.g., traffic flows, number of competitors and complements), and
  • COVID-related measures (local case counts, vaccination rates, restaurant employment levels)


The model results offer insights to performing arts leaders trying to understand and manage supply and demand in the time of COVID. We start with the average price, number of performances, and vaccination rate (47%) in June 2021 to generate a simulation. Figure 2 plots the expected change in average ticket sales associated with a -20% and +20% change in price discount, number of performances, and COVID vaccination rates, with “0” being the June 2021 actual levels.

The key takeaway is that demand for these performing arts organizations in June 2021 was determined more by COVID vaccination rates than by managerially controlled variables associated with how many performances to offer or how to price them. Specifically, ticket sales increase 15% (from 7.75 to 8.9 tickets per household census tract [HHCT]) with a 20% rate of increase in COVID vaccination rates (from 47% to 56% vaccination rate). Ticket sales increase by only 6% with a 20% increase in number of performances and by 5% with a 20% price discount.


We also used the model results to conduct a series of simulations that predict future ticket purchase levels under different assumptions. For example, what would happen to ticket purchase levels if the U.S. had achieved high vaccination rates and COVID cases had remained low? Or what if vaccination rates stall and COVID case rates climb back to high levels as colder weather settles in or a new variant emerges? 


We show three simulations in Figure 3. In each scenario, we hold organizational and household demographic characteristics constant at their June 2021 levels to isolate the effects of COVID. 

  1. The realistic worst-case scenario (Panel A) shows predicted ticket sales reaching 57% of the 4-year high by December 2021:
    • using actual levels of COVID cases, vaccination rates, traffic counts and restaurant employment for July, August, and September 2021, and
    • continuing those trends through the end of the calendar year, with vaccination rates eventually reaching 60% of the total population (including children).
  2. The realistic best-case scenario (Panel B) shows predicted ticket sales reaching 65% of the 4-year high by December 2021:
    • using actual levels of COVID cases, vaccination rates, traffic counts and restaurant employment for July, August, and September 2021, and
    • assuming that COVID cases drop and vaccination rates improve significantly through the end of the calendar year, with vaccination rates reaching 67% of the total population.
  3. The idealized best-case scenario (Panel C) shows predicted ticket sales reaching 79% of the 4-year high by December 2021:
    • assuming vaccination rates escalated from July to December to reach 78%, which is still below the current vaccination rate in Spain (80%),[1] and
    • assuming COVID cases remained at July 2021 levels.

The simulations allow us to infer losses in ticket sales associated with the three scenarios.  The average organization in our sample had approximately $12M in ticket sales in 2019.  For the last 6 months of 2021, the average monthly difference in ticket sales between the realistic worst-case scenario and the realistic best-case scenario is only 4.3%, but the average monthly difference in ticket sales between the realistic worst-case scenario and the idealized best-case scenario is 24%, which represents a $1.5 million loss in annual ticket sales for the average organization in our sample


Extrapolated to the entire nonprofit performing arts industry, the total loss in ticket sales associated with a realistic worst-case scenario vis-à-vis the idealized best-case scenario is nearly $1B. These incremental losses focus only on the recovery period of July-December 2021 and ignore the devastating losses incurred in 2020 and early 2021 (see Figure 1).  We estimate that total losses in the industry attributable to the pandemic through December 2021 likely exceed $3.2B, which dwarfs the $400M allocated as part of the Paycheck Protection Program.[2] We also estimate that lagging vaccination rates cost this industry around $10M per month for every unrealized percentage point in vaccination rates.


The pandemic has changed our lives in many ways, but few if any industries have been more affected than the performing arts. Our model-free data visualization (Figure 1) and econometric model (Figure 2) confirm the dramatic impact that COVID has had. The simulations (Figure 3) calculate the ongoing impact of low vaccination rates and resurging COVID cases and offer insights for performing arts leaders trying to manage supply and demand in the time of COVID.  Quantifying the impact of the pandemic on this industry provides context for better understanding the future for live performances and the important role that vaccinations play.


Limitations and Ongoing Research 

The statistician George Box is often quoted as saying, “All models are wrong, but some are useful.” Our model is surely wrong. This is especially likely in predicting sales levels outside the range of the data used, which currently peaks at a 58% vaccination rate. This suggests that the simulations offered in Figure 2 are more reliable than the simulations offered in Figure 3, especially the idealized, best-case scenario offered in Panel C. Figure 2 shows how the model fits actual data for organizations operating in counties with vaccination rates that ranged from 38-58% in June 2021. Figure 3, on the other hand, extrapolates the model results to scenarios that feature vaccination rates rising to 60-78%. Our model results indicate that the effect of vaccination rates is diminishing, as would be expected for a construct that has a theoretical maximum of 100%. But the rate at which the effect diminishes likely accelerates as the vaccination rate approaches 100%.  Our model cannot capture the extent to which that will occur because we do not currently observe higher vaccination rates. But we will update our model as new data become available, and we will share new results.



[1] accessed 11/05/2021.

[2] See

Expansion of the COVID-19 Benchmark Dashboard, an initiative led by international arts management consultants TRG Arts and U.K. arts data specialists Purple Seven, is supported in part by a grant from the National Endowment for the Arts to SMU DataArts, TRG Arts’ long-time partner in advancing the arts and culture sector.


When the COVID-19 virus began spreading across the U.S., researchers at SMU DataArts responded by integrating datasets and building a framework for predicting ticket purchasing demand. Continually refined for over a year, this framework takes into account ticketing purchases, census data, COVID cases, vaccine rates, restaurant employment, and arts ticket prices to help organizations across the nation predict demand for in-person ticket purchases. Get a behind-the-scenes look at how the model was developed and how early actual results compare with predictions.

Glenn B. Voss, Ph.D., is research director for SMU DataArts and Professor Emeritus from the Cox School of Business at Southern Methodist University. His research – focusing on innovation and organizational learning, customer satisfaction and relationship management, and retail pricing strategies – has appeared in leading academic journals in marketing (e.g., Journal of Marketing, Journal of Marketing Research) and management (e.g., Academy of Management Journal, Organization Science).

Karthik Babu Nattamai Kannan, Ph.D., is an Assistant Professor of Information Technology and Operations Management at Cox School of Business, Southern Methodist University and the Donna Wilhelm Research Fellow at SMU DataArts. He studies how technological innovations are changing how people access the internet, consume digital entertainment and participate in e-commerce platforms. He also studies how retailers and art institutions can leverage mobile location data to improve their service operations. His research interests include location analytics, electronic/mobile commerce, and social media analytics. In his research, he uses empirical methods such as advanced econometrics, machine learning, field/natural experiments, etc., and optimization models to study large-scale datasets.

Data and Research for the Arts Sector

Find More Data-Driven Insights