Impact of COVID-19 on the U.S. Hospitality Sector

Project Overview

This project analyzes the impact of COVID-19's external shocks, such as travel restrictions, case numbers, and mobility trends, on economic outcomes in the U.S. hospitality sector. By examining demand patterns and revenue trends during the pandemic, this study provides insights into the effects of such crises on the industry.

Methodology

We utilized multiple models to assess the impact of COVID-19 on the U.S. hospitality sector:

1. Linear Regression: Served as a solid starting point, demonstrating a strong fit and interpretable relationships between mobility, COVID-related data, and demand. However, residual analysis indicated autocorrelation, necessitating more advanced models.

2. Bass Model & Generalized Bass Model (GBM): Used for modeling diffusion processes. The GBM with an exponential shock successfully captured the demand drop caused by COVID-19 and provided valuable insights into market potential. However, it struggled with seasonality.

3. ARIMA on GBM Residuals: Integrated ARIMA models on the residuals of the GBM to account for seasonality, improving the model fit but slightly overestimating demand.

4. SARIMAX Model: The most effective model for handling seasonality and describing the impact of external variables accounting for mobility in different areas. Significant predictors included:

   • Retail and Recreation
   • Grocery and Pharmacy
   • Parks
   • Workplaces

   The integration of GBM as an X-regressor improved the fit, though some data features remained unmodeled.

Key Findings

• The COVID-19 pandemic had a significant impact on demand in the hospitality sector.
• Mobility indicators played a crucial role in explaining variations in demand and revenue.
• SARIMAX outperformed other models in capturing seasonality and external shocks.
• The integration of diffusion models with time series techniques improved predictive performance but required further refinement.

Conclusion

This study highlights the effectiveness of various modeling approaches in understanding external shocks in complex datasets. It also underscores the importance of incorporating external factors, such as mobility and pandemic-related data, to comprehensively describe the state of the hospitality sector during crises.

Authors

• Akbota Norzhanova
• Aigerim Sagadiyeva
• Aysenur Oya Ozen

Acknowledgments

This project was conducted as part of the Business Financial and Economic Data course. We acknowledge the support of our instructors and the use of publicly available data sources.