Repository logo
 

Exploring the use of credit/debit card transaction data in estimating national park visitor spending: a Mount Rushmore case study

Abstract

Visitor spending refers to trip-related expenditures made by tourists. Estimates of visitor spending are needed for national park economic contribution studies, and are an essential component in evaluating tourism-related economic activity [Thomas et al., 2019, Wilton and Nickerson, 2006]. Traditional methods for estimating visitor spending rely on visitor surveys, which are costly and subject to multiple forms of survey bias [Stynes and White, 2006, Sinclair et al., 2023, Wilton and Nickerson, 2006]. Using Mount Rushmore National Memorial as a case study, I explore the use of granular credit and debit card transaction data to estimate visitor spending without the need for survey data. I use Safegraph Spend as the source of credit and debit card transaction data. I gather transaction data at stores within 100 miles of Mount Rushmore for the period 2019-2023. Combining this data and Mount Rushmore visitation data from the National Park Service, I develop and compare multiple models that use fixed effect regressions to estimate average spending per visit in the Accommodation, Food Service, Retail, and Arts/Entertainment/Recreation sectors. Using results from the best performing model, for 2022, I estimate Mount Rushmore visitor spending to be $41.0 million in the Accommodation sector, $38.4 million in the Food Service sector, $154.8 million in the Retail sector, and $6.4 million in the Arts/Entertainment/Recreation sector. I compare these estimates with NPS survey-based Mount Rushmore visitor spending estimates. I find that model estimates for the Food Service and Retail sectors are statistically indifferent to NPS estimates, however, model estimates for the Accommodation and Arts/Entertainment/Recreation sectors are below NPS estimates. I find several strengths and weaknesses in the credit/debit card transaction models. One strength is the use of observed spending data rather than stated spending data. Another strength is the representation of yearly visitor spending habits. A third strength is the ability to provide measures of estimate precision, like standard errors. One weakness is the inability to identify park-specific visitor spending when other nearby tourist attractions have similar visitation. Another weakness is the failure to account for visitor trip purpose. Additionally, I find that Safegraph Spend underrepresents spending in the Accommodation sector since vacation rental websites (like Airbnb) are not reflected in the data. Looking forward, further research should focus on methodological refinement and the integration of other data sources to improve visitor spending estimation using credit/debit card transaction data.

Description

Rights Access

Subject

Citation

Associated Publications