The effects of 911 on casino revenues: a comparison of Mississippi and Las Vegas.

Moss, Steven E. ; Ryan, Chuck ; Parker, Darrell 等

ABSTRACT

This research compares the effect of 9/11 on casino gaming revenues in Las Vegas and Mississippi. ARIMA models with intervention and transfer functions are used to estimate a time series model for each market. The models show a significant negative downturn in gaming revenues for Las Vegas post 9/11. A similar downturn in gaming revenues in Mississippi is not observed. Air travel is introduced as an explanatory variable for the negative intervention in Las Vegas. This research shows that air travel has significant explanatory power both pre and post 9/11 for Las Vegas gaming revenues. Implications for casino operators are then discussed.

INTRODUCTION

Since September 2001 there has been speculation about the effects of terrorism and changes in the airline industry on various segments of the US economy. In this paper we will show that a statistically significant drop in Las Vegas, Nevada casino gaming has occurred. The methodology used will clearly demonstrate that the decline is not attributable to seasonal shifts or pre-existing trends in gaming revenues. Additionally we will show that there has not been a corresponding drop in Mississippi casino gaming revenues since September 2001.

Having established that Las Vegas gaming revenues have decreased significantly since 9/11, we introduce air travel as a possible explanatory variable. A relatively large percentage of Las Vegas gamblers arrive via commercial aircraft, while most Mississippi gamblers come from adjoining states and do not fly to the casinos.

The airline industry and resorts that rely on airlines maybe some of the hardest hit by the economic effects of 9/11. In the two weeks following 9/11 some 240 conventions cancelled events in Las Vegas (Verhovek & Kaufman, 2001). By October of 2001 forecast national convention revenues had been adjusted downward from 96 billion to 76 billion. The drop in convention revenue is partially attributed to excessive media coverage of the airplane disasters of 9/11 (Barabosa, 2001). Early evidence shows that the decline in air travel may not be short lived. Nationally air travel was down 14.6% in December 2001 versus the prior year. A February, 2002 USA Today survey showed that 43% of respondents reported that they were afraid to fly, almost the same percent as the 44% reported in November of 2001 (Morrison, 2002). Prior to 9/11 only 10% of Americans reported that they were afraid to fly. In a 2003 survey the percentage of Americans afraid to fly was 40% (Fitzpatrick, 2003).

If customers are unwilling to use commercial air service, this reluctance may explain, in part, a downward shift in Las Vegas gaming revenues. If Las Vegas gaming revenues are significantly affected by the general publics willingness to fly while other casino markets do not depend on air travel for their visitors then Las Vegas casinos face a significant threat to their ability to continue to grow and maintain their dominate market share. The same threat that Las Vegas faces may also be an opportunity for casino operators who can operate casinos in regional markets such as Mississippi.

MISSISSIPPI CASINO INDUSTRY

Mississippi ranks as the third largest casino market in the United States, with more than 50 million people visiting the state's casinos each year (Mississippi Gaming Commission, 2003; Meyer-Arendt, 1995). Additionally, operations in the state have been consistently regulated, facilitating analysis over time (Russell, 1997).

Legislation in Mississippi authorizing gaming on navigable waterways was passed in 1990, and the first casino opened in mid-1992. While some have classified Mississippi casinos as a form of riverboat gambling (Roehl, 1994), large facilities with adjoining hotels, restaurants, and entertainment facilities generate most revenues today.

During the period of 1992 to 2002, annual casino gaming revenues increased from $121 million to over $2.7 billion, and the number of properties grew to 30 (Mississippi Tax Commission, 2003). Casino space now exceeds 1.4 million square feet. Of the more than 50 million people who patronize Mississippi's casinos annually, approximately 65% come from Mississippi or adjoining States. Fully 80% come from the Southeastern U.S. (Mississippi Gaming Commission, 2003).

LAS VEGAS CASINO INDUSTRY

Las Vegas was founded as a city in 1905 and the first gambling licenses were issued in 1931, when Nevada legalized gambling. Corporations began to acquire casinos in the 1960's, signaling a makeover in industry strategy ("The history," n.d.). Gaming revenues grew to $7.7 billion in 2000 and 7.6 billion in 2001. In the first twelve months following 9/11 Las Vegas gaming revenues declined $298 million from the prior twelve months.

In 2001 alone, more than 35 million people visited Las Vegas (Las Vegas Convention and Visitors Authority, 2002). Studies show that 86 percent of Las Vegas visitors gamble while there, each with an average gambling budget in excess of $600. Air travel is important to the health of the Las Vegas casino industry. Fully 48% of the city's visitors arrive by air. This importance is illustrated by the fact that while overall interstate and highway traffic was up by an average of 11%, airline passengers were down by 6.7%, and overall visitor volume was down 2.3% during June of 2002 versus the prior year (Las Vegas Convention and Visitors Authority, 2002).

Overseas visitors predominately travel by air to Las Vegas. Japan provides the largest number of overseas visitors. Overall Asian visitor volume at two of the largest casinos is reportedly down by 75% since September 2001 (Berns, 2002). With few other travel options besides air travel these overseas visitors are most likely lost business.

DATA

Two data sets are analyzed in this research, Mississippi gross casino revenues and Las Vegas gross casino revenues. In both cases gross casino revenues represent the amount the casino wins from gaming operations. The gross revenues do not include revenue from other sources such as restaurant or hotel operations. In both series the data is a monthly time series.

The Mississippi casino gross revenues (MSGR) were collected from the Mississippi State Tax Commission, Miscellaneous Tax Bureau, Casino Gross Gaming Revenues reports. The Mississippi series is for all casinos operating in the State of Mississippi. The data cover the time period from August 1992 through July 2002.

The Las Vegas series (LVGR) was collected from the State of Nevada, Gaming Control Board, Tax License Division, Nevada Gaming Revenues and Collections reports. The data come from results of casinos operating in Clark County, NV. Clark County includes the Las Vegas strip, downtown Las Vegas, North Las Vegas, Laughlin, the Boulder strip, and Mesquite. Some smaller areas such as Reno and Lake Tahoe are not included in the data. Clark County accounted for 80.4% of statewide gaming revenues in July 2002, and is the primary casino tourist area in the state. Statistics from November 1996 through September 2002 are used in the study, as these are the public data available from the Nevada State Gaming Control Board as of the date of this research.

To model LVGR monthly enplaned and deplaned airline passengers in Las Vegas is used as an independent variable (AIR). AIR was obtained from the Las Vegas Convention and Visitors Authority. AIR is shown in figure 1. This series is available for December 1996 through September 2002.

[FIGURE 1 OMITTED]

METHODOLOGY

The goal of this research is to estimate a time series model for each series that will allow for the intervention of September 2001 events. A multiplicative decomposition model, shown in formula 1, is used to model each series up to September 2001 (Bowerman & O'Connell, 1993; Moss, Ryan, & Wagoner, 2003). The model is then used to forecast a 95% confidence interval for the months following September 2001.

[Y.sub.t] = T[R.sub.t] x S[N.sub.t] x I[r.sub.t] Formula (1)

The multiplicative decomposition model has the advantage of being relatively easy to fit. However, the trend portion of equation and [R.sup.2] can be misleading due to auto-correlation in the residuals and the limitation of using algebraic manipulations of the period number in the multiple regression.

To verify the multiplicative decomposition model a Box-Jenkins ARIMA model with an intervention factor is also estimated (Vandaele, 1983; Bowerman & O'Connell, 1993). The data for the ARIMA models will be deseasonalized and first differenced to achieve stationary series. The intervention factor will test for the possible effects of 9/11 on the series. A permanent intervention with immediate effect is used. It is assumed any potential impact on the series is immediate as of September 2001. It is also assumed that the impact is permanent.

A transfer function model will then be estimated to show evidence of the effect of an independent variable on Las Vegas gaming revenues. The ARIMA models are estimated using the Box-Jenkins procedure.

ANALYSIS

Both MSGR and LVGR exhibit trends (non-stationary series) and seasonality. The seasonal portion of the multiplicative decomposition is modeled with a ratio to centered moving methodology (Bowerman & O'Connell, 1993; Moss et al., 2003). Seasonal indices are shown in Table 1. The deseasonalized series are shown in Figures 2 and 3.

[FIGURES 2-3 OMITTED]

The seasonal indices for both series have months that deviate significantly from 1.00 supporting the observation of seasonality in the original series. The deseasonalized series, shown in figures 2 and 3, both exhibit trends. The trend equations for the series are shown in formulae 2 and 3. The trend equations are fit using the pre-911 data.

FDMSGR = 36,317,734 + 3,201,921.7*P--1,349.5*[P.sup.2] Formula (2) [R.sup.2] = .951

FDLVGR = 276,000,000 + 3,469,404.5*P Formula (3) [R.sup.2] = .845

The forecast and actual deseasonalized series with 95% confidence limits are shown in figures 4 and 5.

[FIGURES 4-5 OMITTED]

For the DMSGR the actual series stays with the 95% confidence limits after September 2001. For the DLVGR the actual series falls below the 95% confidence limit after September 2001. This result indicates that Las Vegas was hurt significantly by the events of September 2001, whereas Mississippi casinos revenues have not changed.

To verify the multiplicative decomposition model Box-Jenkins ARIMA models with an intervention factor are used. The series are both deseasonalized prior to estimating the ARIMA models. This technique has the advantage of reducing the complexity of the model and maintaining the maximum series length. The deseasonalized data shown in figures 2 and 3 both exhibit non-stationary trends. First differences are used in both series to obtain stationary series.

First differencing results in a stationary series for the DLVGR. For the DMSGR first differences result in a stationary series except for the beginning of the series when casinos first opened in Mississippi. During the initial months of operation the Mississippi casino industry was very small and growing very rapidly. For the initial part of the DMSGR series the first differences do not have the same mean or variance as the remainder of the series. Therefore the first twenty observations from the Mississippi ARIMA model are omitted for model estimation. The ARIMA model for Mississippi with the potential intervention factor is shown in Table 2.

The intervention factor for the Mississippi series, shown in table 2, is not significant. The Ljung-Box Q statistic, auto-correlation, and partial auto-correlation function for the residuals of the model all indicate a white noise residual series. Since the intervention factor is not statistically significant the ARIMA model is re-estimated without the intervention factor. The results are shown in table 3.

The ARIMA model for DMSGR has an [R.sup.2] of 95.7%. The Ljung-Box Q statistic, auto-correlation function, and partial auto-correlation function for the residuals of the model all indicate a white noise residual series.

The ARIMA model for the DLVGR with an intervention factor is shown in table 4.

For the Las Vegas series the intervention factor is statistically significant and negative. This indicates a downward change in revenues after September 2001. The Ljung-Box Q statistic, auto-correlation function, and partial auto-correlation function for the residuals of the model all indicate a white noise residual series. The model was used to forecast the DLVGR; results are shown in figure 6.

[FIGURE 6 OMITTED]

The ARIMA model with intervention in table 4 confirms there is a shift in the series on September 2001. One possible explanation for the shift in the Las Vegas series is air travel. To test this hypothesis an ARIMA model with a transfer function using first differences of deseasonalized AIR as the independent variable was estimated. The model is shown in table 5.

The model shows that air traffic has a significant positive relationship with DLVGR. The Ljung-Box Q statistic, auto-correlation function, and partial auto-correlation function for the residuals of the model all indicate a white noise residual series. Additionally this model correctly predicts positive or negative monthly changes in DLVGR 50 out of 68 months forecast (74%). In the 13 months analyzed post 9/11 the model correctly predicted the direction of the change in DLVGR in 77% of the months. The model in table 5 was used to forecast DLVGR, the results are shown in figure 7.

[FIGURE 7 OMITTED]

To test the robustness of the findings the model in table 5 was estimated using only the data through August 2001. The resulting coefficients change only slightly and all variables are still significant. This verifies that DAIR had a significant relationship with DLVGR prior to September 2001. Testing also included the addition of an intervention to the transfer function model. With DAIR included in the model the intervention is no longer significant.

CONCLUSION

The study shows that Las Vegas gaming revenues have dropped materially since September 11, 2001, while the Mississippi market has not experienced such effects. We believed, a priori, that the downward shift in the underlying pattern of passenger air travel was hurting the Las Vegas market, since a greater percentage of patrons travel there by commercial airline, relative to those visiting Mississippi's casinos. By removing seasonality and trends, we were able to isolate air travel effects and found them to be significantly related to the drop in Vegas revenues.

A recent study using data from 130 domestic airports reports the slump in the airline industry will last at least until 2005 (Banstetter, 2002). Given that, and the relationship between Las Vegas casino revenue and air travel, a recovery does not appear to be forthcoming any time soon. What should Vegas casino/hotel operators and investors do in the meantime?

First, we suggest they concentrate on reducing both monetary and psychological marginal costs of flying to the city. The monetary component could be accomplished by marketing aggressive discount incentives related to meals and the hotel stay. Furthermore, win per square foot per day, or the amount the casino wins from individual patrons (Kilby & Fox, 1998) could be decreased, again coupled with a hard-hitting marketing campaign. Psychological transaction costs can be reduced through new casino alliances with airlines based at smaller, regional airports. Shorter check-in lines and airport commute times should help mitigate the current hassle factor of traveling through major U.S. airports. After all, who truly wishes to stand in a check-in line for an hour or two, after a trip of similar duration to that major airport by auto, in order to fly to Las Vegas: a trip that will also take at least an hour or two, on average?

Second, we suggest that casino/hotel operators and investors consider diversifying their portfolios with regional casinos that do not rely heavily on air travel. For long-term, the health of the industry might be enhanced by increasing retail presence in these smaller, regional markets. The ability of the Mississippi casinos to weather the current economic storm serves as a perfect example. Similar results have been reported in newer, smaller casino markets. In 2002 Missouri casino revenues increased by 18% in February and 11% in March versus the prior year. Missouri increases are attributed to a mild winter, local market, a growing casino industry, and possibly a fear of flying (Wilgoren, 2002).

We realize that the above shifts in tactics and strategy will be expensive. Yet, this expense is mitigated by the consideration that the decrease in airline passengers costs the Las Vegas casino industry around $247 million each year (table 6), and will do so until airline travel begins to recover, perhaps in 2005, perhaps never.

REFERENCES

Barboza, D. (2001, Oct. 23). Rash of No-Shows at Conventions, The New York Times, A16.

Berns, D. (2002, March 29). Las Vegas casino executives: China backlash may hurt influx of high rollers. Knight Ridder Tribune Business News. 1.

Banstetter, T. (2002, August 20). Study predicts long air slump. Knight Ridder Tribune Business News, 1.

Bowerman, B.L. & R.T. O'Connell (1993). Forecasting and time series: An applied approach (3rd ed.). Pacific Grove, CA: Duxbury.

Fitzpatrick, M. (2003). Fear of Flying. Lancet, 361(9353), 268.

Kilby, J. & J. Fox (1998). Casino operations management. New York: Wiley.

Las Vegas Convention and Visitors Authority. (2002). Executive summary of tourism indicators. Retrieved January 25, 2003, from http://www.vegasfreedom.com

Las Vegas Convention and Visitors Authority. (2002). Frequently asked research questions. Retrieved January 25, 2003, from http://www.lasvegas24hrs.com/ gen_resfaq.asp

Las Vegas Convention and Visitors Authority. (2002). 2001 Las Vegas visitor profile study. Retrieved January 25, 2003, from http:// www.lasvegas24hrs.com/gen_vispro.asp

Las Vegas Convention and Visitors Authority. (2002). Visitor statistics: 1970 to present. Retrieved January 25, 2003, from http://www.lasvegas24hrs.com/gen_vstat.asp

Meyer-Arendt, K.J. (1995). Casino gaming in Mississippi: Location, location, location. Economic Development Review, 13(4), 27.

Mississippi State Tax Commission, Miscellaneous Tax Bureau. (n.d.). Casino gross gaming revenues reports. Retrieved January 25, 2003, from http://www.mstc.state.ms. us/taxareas/misc/gaming/stats/GamingGrossRevenues.pdf

Mississippi Gaming Commission. (2003). Public Information--Quarterly Survey Information. Retrieved January 25, 2003, from http://www.mstc.state.ms.

Morrison, B. (2002, Feb, 14) Americans still fear flying, poll finds; Though majority say changes have improved security, USA Today, A1.

Moss, S.E., C. Ryan & C. Wagoner. (2003). Forecasting casino revenues: An empirical test of Butler's model. Journal of Travel Research. 41(4), 393-399.

State of Nevada, Gaming Control Board, Tax License Division. (n.d.). Nevada gaming revenues and collections report. (n.d.). Retrieved January 25, 2003, from http://gaming.state.nv.us/mrrindex.htm

The History of Las Vegas. (n.d.). Retrieved January 25, 2003 from www.ci.las-vegas.nv.us

Roehl, W.S. (1994). Gambling as a tourist attraction: Trends and issues for the 21st century. In Seaton, A. V. et al. (Eds.), Tourism: The State of the Art (pp. 156-168). Chichester, West Sussex, UK: John Wiley & Sons.

Russell, K. (1997). Survival of fittest emerges as theme of state casino market. Mississippi Business Journal, 19(18), 1-2.

Vandaele, W. (1983). Applied time series and Box-Jenkins models. San Diego: Academic Press.

Verhovek, S.H. and L. Kaufman (2001, Sep. 20). America's Fear of Flying Has Devastating Effect On Tourist Business, The New York Times, B3.

Wilgoren, J. (2001, May 20). Midwest Towns Feel Gambling Is a Sure Thing, The New York Times, A1.

Steven E. Moss, Georgia Southern University Chuck Ryan, Georgia College & State University Darrell Parker, Georgia Southern University

Table 1: Seasonal Indices

 MS LV

January .987 1.153
February .987 .935
March 1.087 1.042
April 1.006 .971
May 1.016 1.022
June .994 .911
July 1.125 .980
August 1.041 .986
September .966 .972
October .956 1.032
November .937 .993
December .897 1.003

Table 2: Mississippi ARIMA model w/ Intervention Diff=1

Variable Coeff Std Error T-Stat Signif

CONSTANT 1,313,569.515 373,652.282 3.515 0.001
AR{1 -0.645 0.090 -7.174 0.000
AR{2 -0.478 0.091 -5.255 0.000
INTERVENTION{0} -946,034.946 6,305,328.845 -0.150 0.881

[R.sup.2] = 0.957

Ljung-Box Q-Statistics: Q(12)=10.228 Significance Level = 0.596

Table 3: Mississippi ARIMA model w/o Intervention Diff=1

Variable Coeff Std Error T-Stat Signif

CONSTANT 1,304,116.052 366,598.780 3.557 0.001
AR{1} -0.644 0.089 -7.210 0.000
AR{2} -0.479 0.090 -5.317 0.000

[R.sup.2] = 0.957

Ljung-Box Q-Statistics: Q(12) = 10.329. Significance Level = 0.587

Table 4: DLVGR w/Intervention Diff=1

Variable Coeff Std Error T-Stat Signif

CONSTANT 3,481,581.38 1,585,057.93 2.197 0.032
AR{1} -0.84 0.13 -6.577 0.000
AR{2} -0.28 0.12 -2.279 0.026
INTERVENTION{0} -62,903,771.98 20,394,911.89 -3.084 0.003

[R.sup.2] = 0.801

Ljung-Box Q-Statistics: Q(12) = 18.405. Significance Level = 0.104

Table 5: ARIMA w/ Transfer Function

Variable Coeff Std Error T-Stat Signif

CONSTANT 1,836,640.066 1,411,027.502 1.302 0.198
AR{1} -0.817 0.104 -7.854 0.000
AR{2} -0.352 0.096 -3.656 0.001
DAIR{0} 76.033 18.202 4.177 0.000

[R.sup.2] = 0.826

Ljung-Box Q-Statistics: Q(12) = 15.380. Significance Level = 0.221

Table 6: Estimate of revenues lost owing to decreased air travel
(in U.S. dollars).

2000 gaming revenues $7,700,000,000
% travel by air x .480
% decrease in air travel x .067
Yearly revenues lost $247,632,000