Hotel tax receipts and the ‘Midnight in the Garden of Good and Evil’: a time series intervention seasonal ARIMA model with time-varying variance

Abstract

This study examines the influence of the release of a best-selling book and movie, Midnight in the Garden of Good and Evil, set in Savannah, Georgia on local tourism demand. Tourism demand is proxied by revenue collected from an ad valorem hotel room tax in Savannah. The hotel tax revenue series is first modelled as a seasonal ARIMA model with three intervention variables: an index variable to capture the influence of the best-selling book and two dummy variables to represent the impact of the 9/11 terrorist attacks and hurricane Floyd. The presence of time-varying variance in the residuals is captured through an ARCH model. The results indicate that the book index had a positive and significant impact on hotel tax receipts, while the dummy variables for the terrorist attacks of 9/11 and hurricane Floyd were each negative with only the dummy variable for hurricane Floyd marginally significant.

Notes

¹ See study by McGrath and Toma (2004).

² In 2004, the share of overnight accommodations in a hotel or motel was 88% (McGrath and Toma, 2004).

³ The highest value of this variable (BOOK) is one, while the lowest is 0.04 (equals 1/25), since books drop off the list if their rank exceeds 25. This mitigates the problem of missing data in ranks higher than 25 and the problem of relative comparisons of the value zero (not ranked), one (best rank) and missing values for unknown ranks >25.

⁴ Blink et al. (2006) examine the adverse impact of the September 11th terrorist attacks on the airline industry.

⁵ The MA(1) and MA(4) terms satisfy the invertibility conditions, |θ₁| < 1 and |Θ₁| < 1, respectively.

Table 1. Model results 1983:1–2006:2

⁶ As discussed in Bollerslev and Woolridge (1992), the model is estimated under the assumption that the errors are conditionally normally distributed with covariances and standard errors computed via quasi-maximum likelihood.