Fear and managing in Las Vegas: an analysis of the effects of September 11, 2001, on Las Vegas...

This article provides an analysis of the impacts of the events of September 11, 2001, on certain gaming business volume on the Las Vegas Strip. Almost immediately after these events, management teams commenced with layoffs and project postponements, in anticipation of a long and painful recovery

period. The authors' intervention analysis, however, reveals an abrupt decline lasting five months, indicating a relatively rapid recovery. Interestingly, this type of recovery appears to be similar to that which has been observed in other literature examining previous terrorist attacks. In the concluding discussion, the authors highlight the lessons of both history and empirical inquiry in hopes of helping managers understand the dynamics of terrorism impacts on the hospitality industry in general and on the gaming industry in particular.

Keywords: travel and tourism; Las Vegas Strip; terrorism; slot coin-in

Contrary to expectations that the gaming business would face a lengthy disruption after the terrorism attacks of September 11, 2001, gaming revenue for the Las Vegas Strip had largely recovered by the following January. An intervention analysis (ARIMA) of slot coin-in, which is the purest gauge of gaming activity, recorded an abrupt drop in revenues in September. However, the analysis found that business started picking up just as suddenly about five months after the attacks, although full recovery took two years. Concerned by the sudden reduction in business in September, the Strip's managers laid off employees and postponed capital projects during the falloff period. The hasty cutbacks by managers did not take into account the possibility of a quick recovery. The pattern experienced by Las Vegas echoes that of other destinations beset by terrorist activity, which show relatively speedy recovery in the absence of further terrorist attacks.

**********

By some estimates, the economic costs to the United States of the terrorist attacks of September 11, 2001, included more than $14.0 billion for private enterprises and in excess of $2.2 billion for federal, state, and local governments. Meanwhile, estimates of cleanup and rescue costs exceeded $11.0 billion (Center for Contemporary Conflict n.d.), and estimates of the largest single cost associated with the events of 9/11, insurance, came in at around $75 billion (Burritt 2001-2002).

Understandably, observers of hospitality markets feared the worst, with many industry analysts and executives predicting a lengthy downturn. Bill Schmidt, a gaming industry analyst with CIBC World Markets in New York, summed up the concerns of many when he said, "You have to anticipate the worst ... that this will change the psychology of people traveling for a long time" (Financial Times 2001).

Indeed, the tourism business plunged after 9/11, but as we discuss in this article, other economic trends were also at work. First, we examine the literature to determine how various tourism markets have recovered in periods following significant terrorist attacks. We believe that although the events of 9/11 were unprecedented, that does not render historical analyses meaningless; in fact, such analyses might prove especially important to counter short-term views driven by fear or hysteria.

Next, this article estimates the length and intensity of the downturn experienced by Las Vegas Strip casinos through an intervention analysis based on an autoregressive integrated moving average (ARIMA) model. For this analysis, we examine slot coin-in, the purest gaming volume indicator that managers have at their disposal. We first develop a model that predicts the expected Las Vegas Strip gaming volume (absent the attack) for the end of 2001, and then we compare the predicted values with the actual gaming volume.

With this numerical analysis as a backdrop, we examine the gaming managers' decisions made in the immediate aftermath of 9/11, which included layoffs and project delays and cancellations that damaged employee morale and interfered with needed supply availability. Given that analysts have raised legitimate questions about the rationality of these decision-making processes (Wall Street Journal 2002), this article provides historical and empirical foundations for well-thought-out management decisions--in particular, decisions made during trying times.

Historical Effects of Terrorism on Tourism

Examining the research concerning the effects of terrorist attacks on tourism, we note a pattern of abrupt reduction, followed by a recovery that occurs in relatively short order. Recovery seems contingent on an absence of subsequent attacks, however, and recovery seems speediest in regions not directly affected by attacks. Brady and Widdows (1998) examined the decrease in travel from the United States to Europe during summer 1986, a season that followed a rash of terrorist-related activity. Interestingly, while the early summer months of 1986 saw a significant downturn, a noticeable recovery took place before summer's end (Brady and Widdows 1998). (See Exhibit 1 for a listing of terrorist attacks over the prior twenty years.)

In a separate review of the impacts of terrorism on tourism in the mid-1980s, D'Amore and Anuza (1986) noted that prior to 1985, the effects of terrorist attacks on tourism tended to last for only a few months. Following the onslaught of attacks in 1985 and 1986, however, affected regions experienced recoveries that were somewhat more dilatory, though the authors noted that these slowdowns were often contained to areas associated with terrorism. The authors added that the longer recoveries experienced in 1986 could be attributed to so many events occurring within such a short time period.

In a sophisticated analysis similar to our work in this article, Coshall (2003) empirically examined travel-related recovery from multiple terrorist incidents using an intervention analysis. This kind of analysis is especially useful because it provides specific assessments of not only the significance of an attack, but also its effects. Regarding the effects, Coshall categorizes terrorist interventions into four basic types, as displayed in Exhibit 2. Those are permanent and abrupt, temporary and abrupt, permanent and gradual, and temporary and gradual.

Based on this typology, Coshall (2003) parsed the effects of three terrorism-related events (namely, the U.S. bombing of Libya in April 1986; the bombing of Pan Am flight 103 over Lockerbie, Scotland, in December 1988; and the Persian Gulf War of 1990 to 1991) on consumer air traffic between the United Kingdom and the United States (among other destinations). Intervention analyses determined that each of these three events created abrupt, but temporary interference in air travel between the United Kingdom and United States (Coshall 2003).

[GRAPHIC 2 OMITTED]

In a similar study, Brady and Widdows (1998) found that U.S. air travel to Europe experienced a similar abrupt, temporary diminishment following the wave of global terrorist events occurring in 1985 and 1986. Overall, although these studies are limited to a specific (albeit substantial) period of terrorist-related activity, they do reveal a consistent pattern of relatively sudden changes: a rapid fall-off in demand followed by a relatively rapid recovery to preattack levels--especially in areas unaffected by the attack.

9/11's Effects on Travel and Tourism

The events of 9/11 also precipitated an abrupt, severe reduction in travel and tourism in the United States. The specific repercussions reported in the three weeks following the attacks included the following:

1. Downtown Chicago restaurants experienced business declines of 45 to 60 percent.

2. Orlando conventioneers cancelled 345 meetings comprising a projected 55,500 participants in the months of September and October 2001.

3. Hotel occupancy in Washington, D.C., San Francisco, and Oahu dropped by 50 percent compared to prior year periods in the weeks following the attacks (Walkup 2001).

In addition, experts anticipated immediate-term reductions in air travel resulting from reduced consumer confidence, safety concerns, and the threat of future attacks (Helgerson 2002; Joint Economic Committee n.d.). Meanwhile, as was the case with Orlando, large convention and business travel cities experienced significant cancellations for conventions and other events occurring immediately after 9/11 (Joint Economic Committee n.d.). Some experts suggested that business travel would experience a prolonged downturn as technologies such as teleconferencing and virtual offices took on a new relevance (Fetto et al. 2001). Ominously, many predicted that travel and other tourism costs would skyrocket due to increased insurance and security costs (Center for Contemporary Conflict n.d.).

In the face of these negative prognostications alarmingly voiced by an understandably concerned punditry, voices of sober optimism also attempted to provide some perspective. Most notably, Alan Greenspan, then chairman of the Federal Reserve Board, stated immediately following the attacks, "We must not lose sight of our longer-run prospects, which have not been significantly diminished by these terrible events" (Greenspan 2001). Ultimately, Greenspan's words would prove not only comforting, but also prescient--at least as they pertained to the subject of our study.

9/11's Effects on Las Vegas

Prior to September 11, 2001, the Nevada gaming industry was ensnared in a slowdown that was also affecting the broader national economy. Furthermore, the state's financial prospects were also diminished by the lack of new properties opening in 2000 and 2001 and by competition from Native American casinos in California, its primary feeder market (Las Vegas Review-Journal 2001a). These factors are no doubt reflected in the financial data from the period immediately prior to 9/11. For the fiscal year ending June 30, 2001, Nevada casinos reported a total win of $9.67 billion, a slight increase over the prior year's win of $9.46 billion, but the smallest percentage gain in four years.

Other competitive threats loomed for Las Vegas during this period. In a trend dating back to 1977, the number of states allowing legalized casino gambling grew from one (Nevada) to thirty-six. As of year-end 2003, there were 443 commercial casinos in eleven states generating revenue of $27.0 billion (American Gaming Association n.d.). In addition, there were 377 American Indian casinos in twenty-eight states (including some that already had casinos), generating revenue of $16.7 billion (National Indian Gaming Commission n.d.).

Nevada's (and more specifically, Las Vegas's) decreasing contribution to the national gambling landscape had meaningful implications for Las Vegas gaming operators after 9/11. The availability of nearby (i.e., drive-to) gaming options meant that would-be gamblers did not have to fly to gamble, and they could avoid Las Vegas, which itself was occasionally cited as a high-profile terrorist target (Etzel 2001). The fact that gamblers had so many choices other than Las Vegas confounded the recovery prospects of the Strip gaming industry.

Given this landscape, it is not surprising that the impact of 9/11 was quickly felt on the Strip (Joint Economic Committee n.d.). The reduction in tourists was immediate, with hotel occupancy rates in Clark County falling by more than 12 percent in September 2001 and 8 percent in October 2001 (compared with prior year periods) (Nevada Gaming Control Board 2001, 2002). This drop-off was also observable in the decline in arrivals at Las Vegas's McCarran Airport, which experienced year-over-year declines in passenger arrivals of 22.2 percent in September 2001 and 9.2 percent in October 2001 (McCarran International Airport n.d.).

Las Vegas gaming operators were quick to respond, with initial views projecting a recovery that would take at least a few quarters. Tom Gallagher, CEO of Park Place Entertainment, painted the immediate future in bleak shades: "Clearly, we're in for some rugged times over the next few months. The recovery will be gradual" (Betas 2001a).

Gallagher was hardly alone in his prognosis, as reflected in the strategies employed by gaming executives to increase demand and decrease costs. On the demand side, for example, MGM Mirage sought to attract customers by lowering room rates by more than 30 percent in the weeks following the attacks. At the company's flagship property, Bellagio, rooms were discounted to as low as $129 per night, down from the previous year's average daily rate (ADR) of $300 per night (Etzel 2001). Other large Strip properties established similar discounts, with the Venetian reducing room rates to as low as $152 per night, from an average of $229 in the weeks before the attacks (Betas 2001a).

Meanwhile, to reduce expenses, the primary strategy used by executives was to lay off or furlough employees. The results were dramatic: in the two weeks following the attacks, an estimated twelve to fifteen thousand employees were laid off from Las Vegas casinos (Las Vegas Review-Journal 2001b). MGM Mirage (three thousand employees), Park Place Entertainment (fifteen hundred employees), and Mandalay Resort Group (forty-five hundred employees) accounted for a large portion of this total (Las Vegas Review-Journal 2001b).

In addition to layoffs, large Strip operators also decided to delay or dramatically scale back major projects. Among the most sizable of these projects were the following:

1. Mandalay Resort Group announced a delay in construction of its 1.8-million-square-foot convention center to be located next to Mandalay Bay Resort (Betas 2001b).

2. Park Place Entertainment announced plans to indefinitely delay construction of a $475 million hotel tower at its Las Vegas Caesars Palace (Berns 2001b).

3. The Venetian Resort announced a construction delay in the property's one-thousand-room tower expansion (Las Vegas Review-Journal 2001c).

Against this backdrop of layoffs and delays in capital investments, let us examine what actually happened to gaming revenues in Las Vegas after 9/11.

Advantages of ARIMA

Researchers have employed various methodologies to forecast gaming revenues. Cargill and Eadington (1978) completed some of the first published research in forecasting gaming volume in an effort to assist governments in estimating gaming-related taxes. Using a variety of methods, the authors attempted to develop forecasting models for Washoe County (including Reno-Sparks), Clark County (including Las Vegas), and Douglas County (including South Lake Tahoe). The techniques used included econometric models, naive models, and an ARIMA model. Of these methods, the ARIMA model was deemed the most appropriate, based on the value of time-series models in the presence of both identifiable seasonal and systematic trends. Analysis using econometric and naive forecasting methods resulted in similar but less accurate forecasting results (Cargill and Eadington 1978).

Shonkwiler (1992) followed up the work of Cargill and Eadington (1978) in predicting taxable gaming revenues in Nevada. While promoting the use of an ARIMA model due to the assumed random nature of the data, Shonkwiler noted that seasonality in the data could be modeled using quarterly dummy variables. In explaining his concerns about the use of econometric models, Shonkwiler cited these models' inability to incorporate the overall trends that have occurred in gaming over time (notably, widespread legalization). Shonkwiler also claimed that "since the gross taxable gaming revenue series represents a mixture of other quantifiable measures and a conglomeration of different types of economic behavior, any explanation in terms of behavioral equations would have nowhere near the parsimony of the model adopted" (p. 248).

In assessing the effects of regulatory changes in Atlantic City, Nichols (1998) also used an intervention analysis based on an ARIMA model. That study attempted to empirically test the effects of increased operating hours and expanded slot machine floor space in Atlantic City casinos, beginning in 1991. Methodologically speaking, this study is relevant to an examination of the effects of 9/11 given the nature of deregulation. Similar to September 11, deregulation requires an analysis of a time series where a single inflection point can be identified.

In sum, ARIMA models provide numerous advantages in time-series analysis, especially when analyzing gaming data. A particularly important advantage of ARIMA models when dealing with gaming data is their ability to account for seasonal and systematic trends (Cargill and Eadington 1978). Also, as mentioned by Cargill and Eadington (1978), the predictive ability of ARIMA models is superior to that of other analyses (e.g., econometric, naive). Finally, Shonkwiler (1992) noted the parsimony of ARIMA models relative to other forecasting tools in predicting gaming revenue. For these reasons, we chose ARIMA models for our analyses in this article.

Data Collection

We obtained data for this analysis from monthly information released in the Nevada Gaming Control Board Gaming Revenue Reports from January 1990 through November 2004. As our indicator in the 179 monthly periods examined, we chose slot machine coin-in, or the total dollar amount wagered in slot machines during play. As we mentioned above, slot coin-in is the purest available gaming volume indicator (Lucas and Santos 2003; Lucas, Dunn, and Singh 2005; Lucas and Dunn 2005; Lucas, Dunn, and Kharitonova 2006; Lucas and Bowen 2002), and it is also the only accurate volume indicator collected by casinos. Moreover, slot coin-in is purer than revenue as it does not factor in luck or short-term volatility (Kilby and Fox 2005). Another strong mason for using this indicator of overall gaming demand on the Las Vegas Strip lies in its increasing importance in the overall gaming mix (see Exhibit 3). In contrast, all other gaming volume measurements contain significant flaws. For instance, most casinos do not or cannot track the amount wagered in table games, and as a result, making accurate computations of theoretical win for table games is exceedingly difficult or impossible. Table games drop is a crude volume indicator at best, because it represents buy-in and not wagering volume (and furthermore, these two variables are not necessarily related).

ARIMA Analysis

To identify gross features of the time series, the data were first seasonally decomposed using the ratio-to-moving average method, also known as Census Method I. As noted from the plots in Exhibit 4, the data exhibited a noteworthy upward trend as well as a strong seasonal component.

Next, the autocorrelation function (ACF) and partial autocorrelation function (PACF) of the residuals were plotted to determine other components of the ARIMA model. An examination of these plots (in Exhibit 5) shows the presence of a moving average (MA) component of 1, as indicated by the significant peak at lag 1. The presence of significant lags at and surrounding lags 12 and 24 also indicated the presence of a seasonal autoregressive (AR) component of 1.

The data were next fitted to an ARIMA model of the form (0,1,1) x (1,0,0). To determine goodness of fit, this model was evaluated using a number of diagnostic checks. First, t-values for both the moving average and seasonal autoregressive terms were determined. Both components had t-values that were significant at the .05 level. Next, ACF and PACF residual plots were reviewed (see Exhibit 6). These plots show the elimination of a significant peak at lag 1, although significant coefficients remain at lags 13 and 17. Box-Ljung statistics were also examined through the first quartile of the data. Exhibit 7 shows significance levels above the .05 level through the first forty-five data points.

Finally, a p-p probability plot of the residuals was examined. Exhibit 8 indicates that residual errors followed a normal distribution. The diagnostic checks conducted indicated that the model is a good initial fit to the data.

We also wanted to account for the dramatic growth periods that have occurred on the Las Vegas Strip since 1989 (detailed in Exhibit 9). The Strip went through two significant expansions during the 1990s, in 1993 and 1998 to 1999. The first expansion, occurring in late 1993, saw the openings of Luxor, Treasure Island, and the MGM Grand, which added approximately 10,400 rooms and 355,000 square feet of casino space to the Las Vegas Strip. The second expansion, which began in late 1998 with the opening of Bellagio, concluded in late 1999 with the opening of Paris Las Vegas. These properties increased the number of rooms by 12,600 and casino space by 495,000 square feet.

We hypothesized that the sudden increase in capacity from these openings changed the nature of the demand on the Las Vegas Strip, primarily by expanding the possible volume. To test this hypothesis, dummy variables were added to the existing ARIMA model in the months surrounding property openings in 1993 and 1998 to 1999. In each case, the months up to and including an opening were represented by the variable 0, while all periods following an opening contained the variable 1. The first dummy variable was included for all months after December 1993, accounting for the opening of Luxor, Treasure Island, and the MGM Grand. The second dummy variable was included for all months after March 1999, accounting for the opening of Bellagio and Mandalay Bay. The openings of the Venetian and Paris Las Vegas may have occurred far enough apart in time from the other hotel openings that their effects on demand were insignificant. Additionally, the high-end gaming focus of the Bellagio and Mandalay Bay may have accounted for a dramatic increase in gaming revenues on the Strip.

[GRAPHIC 4 OMITTED]

[GRAPHIC 5 OMITTED]

Again, diagnostic checks were completed on the new ARIMA model containing dummy variables accounting for the two sets of casino openings, t-values for all components were significant at the .05 level. ACF and PACF residual plots (see Exhibit 10) show improved results relative to those found in the original ARIMA model, with the only significant lag remaining at lag 17. Box-Ljung statistics were examined through the first quartile of the data. Exhibit 11 shows these values to be significant at the .05 level through the first forty-five data points.

[GRAPHIC 6 OMITTED]

Finally, we conducted a normal probability plot of the residual errors from the ARIMA model with dummy variables. Exhibit 12 indicates that the plot followed a normal distribution.

Using data from January 1990 through August 2001, the ARIMA model with dummy variables predicted slot coin-in for September 2001 and the months following the aftermath. Diagnostic checks indicated that the base ARIMA model with dummy variables was a good fit to this subset of the data. Exhibit 13 compares the predicted values to actual values for slot coin-in in the months following August 2001.

Results: Bounce Back with a Long Tail

The results of these projections indicated that the effects of 9/11 lasted from September 2001 through January 2002, with the predicted value for February 2002 resulting in a value slightly above the projected lower confidence limit. To determine what, if any, the lasting effects of 9/11 were, a dummy variable was then created to account for the five-month affected period.

Using data from January 1990 through January 2002, the ARIMA model with the new dummy variable predicted slot coin-in beyond January 2002. Diagnostic checks run indicated that the base ARIMA model with the new dummy variable was a good fit to this subset of the data. Exhibit 14 shows the predicted values compared to actual values for slot coin-in in the months following January 2002.

[GRAPHIC 8 OMITTED]

[GRAPHIC 10 OMITTED]

From February 2002 through February 2004, the predicted values for slot coin-in fall in the lower half of the predicted range. It is not until March 2003 that the actual value exceeds the mean predicted value. These results indicated that the bulk of the Las Vegas Strip's recovery occurred relatively quickly, though there were some lingering effects that lasted for approximately two years.

Management Implications and Further Research

The terrorist-related research that we reviewed indicated that once terrorist activity stops tourism and travel markets tend to recover relatively quickly from their initial rapid falloff in volume (Brady and Widdows 1998; Coshall 2003). These findings appear to be consistent with our findings in Las Vegas in relation to the 9/11 attacks. We note that some of the management decisions to lay off staff and postpone capital projects in Las Vegas occurred during the fall-off period after the 9/11 attacks. Our analysis indicates that managers might better have postponed these decisions with the expectation of a prompt recovery, but the managers could not have been aware of the analysis that we just conducted. Instead, as we indicated above, the data at hand incorrectly indicated an extended bleak period.

[FIGURE 12 OMITTED]

[FIGURE 13 OMITTED]

[FIGURE 14 OMITTED]

The consequences of the decisions were not positive. The sudden layoffs generated significant ill will among employees. This ill will was exacerbated following the disclosure of significant bonus awards to executives at many of the larger companies (Wall Street Journal 2002). As one anonymous Strip executive stated, "There was a perception among the labor team that some of the general managers saw their bonuses growing wings and flying away, and before they knew what was happening, they just panicked and laid people off" (Wall Street Journal 2002). Although some reduction in workforce may have been warranted given the significant slowdown in business, the severe corporate reaction appeared to aggravate an already difficult situation.

We also believe that the indefinite postponement of the capital projects must have interfered with operators' ability to meet the increased demand that followed the resumption of travel activity.

We acknowledge the challenge of the unenviable task of making thoughtful decisions during unthinkably difficult times. However, our study points to the need to set aside understandable fears that inform decision making during difficult times. Another time an unexpected event occurs, managers would be well advised to heed the words of the sociologist Barry Glassner (1999), who devoted an entire research monograph to "The Culture of Fear":

   We had better learn to doubt our inflated fears before they destroy
   us. Valid fears have their place; they cue us to danger. False and
   over-drawn fears only cause hardship. Even concerns about real
   dangers, when blown out of proportion, do demonstrable harm. We all
   pay one of the costs of panics: huge sums of money go to waste.
   (Pp. xvi-xvii)

Limitations of this study might be addressed through additional research. Although slot coin-in represents the strongest indicator of gaming volume, value exists in comparing the recovery of various games (e.g., blackjack, craps, and Baccarat). For example, Gu (1999) noted the importance of Asian customers' contribution to Baccarat gaming volume. Given the drop-off in air traffic to Las Vegas in the months following the events of 9/11, games characterized by high-end play (notably, Baccarat) may have recovered more slowly than did slot machine gaming volume. Future research might evaluate these areas to provide a more comprehensive portrait of recovery.

Although the Strip constitutes the bulk of the Las Vegas gaming market, other areas of the city (e.g., the Boulder Strip, Downtown, North Las Vegas) contribute substantially to the city's overall gaming volume. In contrast to this study's focus on tourist-based business, studies of these areas may provide insight into the recovery of various locals' markets.

Conclusion

Generally speaking, gaming is not necessarily more vulnerable to interruptions than are other hospitality sectors, but Las Vegas makes an excellent case for study. This market is the largest tourist destination in the country and also has more complete published data than most other destinations. As opposed to measuring indicators such as visitor counts or air traffic, with gaming we can actually measure pure business demand through certain gaming volumes.

Our study, based on slot machine coin-in volume, supports previous terrorist-related tourism research showing that markets typically recover fairly quickly after major terrorist attacks, provided no further attacks occur. These results also provide an interesting contrast to the expectations of management and analysts, who predicted a more protracted recovery.

Should another attack occur, management leaders may want to prepare for a more rapid recovery. At a minimum, managers should be more careful in approaching long-term decisions. Given that nothing like the 9/11 attacks had previously occurred in the United States, it is easy to play Monday morning quarterback, but overall we seem to have found a common theme: business levels frequently return back to normal more quickly than the masses (or the managers) fear.

References

American Gaming Association. n.d. 2004 state of the states: The AGA survey of casino entertainment. http://www.americangaming.org/survey/index.cfm.

Betas, David. 2001a. Gaming's recovery to take time: Las Vegas casino executives optimistic that struggling industry will recover. Las Vegas Review-Journal, October 3. http://www.reviewjournal.com/lvrj_home/2001/Oct-03-Wed-2001 /business/17135262.html.

--. 2001b. Mandalay delays convention center project, other casino companies may change plans. Las Vegas Review-Journal, September 22. http://www.reviewjournal.com/lvrj_home/2001/Sep-22-Sat-2001/business/ 17059234.html.

Brady, John, and Richard Widdows. 1998. The impact of world events on travel to Europe during the summer of 1986. Journal of Travel Research 26 (3): 8-10.

Burritt, Chase. 2001-2002. The road to recovery: A look at the lodging industry, post-September 11. Real Estate Issues 26 (4): 5-18.

Cargill, Thomas, and William Eadington. 1978. Nevada's gaming revenues: Time characteristics and forecasting. Management Science 24 (4): 1221-30.

Center for Contemporary Conflict. n.d. Economic costs to the United States stemming from the 9/11 attacks, www .ccc.nps.navy.mil/rsepResources/si/aug02/homeland .asp.

Coshall, John. 2003. The threat of terrorism as an intervention on international travel flows. Journal of Travel Research 42 (1): 4-12.

D'Amore, Louis, and Teresa Anuza. 1986. International terrorism: Implications and challenge for global tourism. Business Quarterly 51 (3): 20-29.

Etzel, Barbara. 2001. Recession-proof no longer Las Vegas hit hard by events; riverboat casinos fare much better. Investment Dealers' Digest 67 (38): 1.

Fetto, John, Pamela Paul, Suzanne Riss, Alison Wellner, David Whelan, Sandra Yin, et al. 2001. What's next? 9.11.01. American Demographics 23 (10): 34-46.

Financial Times. 2001. Vegas counts cost of attacks: Casinos are reeling from a fall in visitors since September 11. October 8.

Glassner, Barry. 1999. The culture of fear: Why Americans are afraid of the wrong things. New York: Basic Books.

Greenspan, Alan. 2001. Address by Alan Greenspan, chairman of the Federal Reserve Board delivered before the Committee on Banking, Housing, and Urban Affairs, U.S. Senate, Washington, D.C., September 20, 2001. Vital Speeches of the Day 67 (24): 756-57.

Gu, Zheng. 1999. The impact of the Asian financial crisis on Asian gaming activities: An examination of Las Vegas strip casino drops. Current Issues in Tourism 2 (4): 354-65.

Helgerson, John. 2002. Global trends and the implications of the 11 September attacks. Presented at the Army War College, Carlisle, PA, January 22.

Joint Economic Committee. n.d. Background material on the potential economic impacts of the terrorist attacks, September 21, 2001. http://jec.senate.gov/democrats/ Documents/Reports/economicaftermathq&aupdated .pdf.

Kilby, Jim, and Jim Fox. 2005. Casino operations management. New York: John Wiley.

Las Vegas Review-Journal. 2001 a. Casinos experience revenue slowdown. August 11, sec. 1A.

--. 2001b. 4,500 cut by resort group. September 28, sec. 1D

---. 2001c. Staff layoff counts climb for casinos. September 26, sec. 1D.

Lucas, Anthony E, and John T. Bowen. 2002. Measuring the effectiveness of casino promotions. International Journal of Hospitality Management 21 (2): 189-202.

Lucas, Anthony E, and William T. Dunn. 2005. Estimating the effects of micro-location variables and game characteristics on slot machine volume: A performance-potential model. Journal of Hospitality and Tourism Research 29 (2): 170-93.

Lucas, Anthony E, William T. Dunn, and Anna Kharitonova. 2006. Estimating the indirect gaming contribution of bingo rooms. Gaming Research and Review Journal 10 (2): 39-54.

Lucas, Anthony E, William T. Dunn, and Ashok K. Singh. 2005. Estimating the short-term effect of free-play offers in a Las Vegas hotel casino. Journal of Travel and Tourism Marketing 18 (2): 53-68.

Lucas, Anthony E, and Jocelina Santos. 2003. Measuring the effect of casino-operated restaurant volume on slot machine business volume: An exploratory study. Journal of Hospitality and Tourism Research 27 (1): 101-17.

McCarran International Airport. n.d. Passengers (domestic & international combined), http://www.mccarran.com/04_04_stats_01.asp.

National Indian Gaming Commission. n.d. Tribal gaming revenues, http://www.nigc.gov/nigc/nigcControl?option= TRIBAL_REVENUE.

Nevada Gaming Control Board. 1990-2004. Gaming revenue report. Las Vegas: Nevada Gaming Control Board.

--. 2001. Nevada gaming abstract. Las Vegas: Nevada Gaming Control Board.

--. 2002. Nevada gaming abstract. Las Vegas: Nevada Gaming Control Board.

Nichols, Mark. 1998. The impact of deregulation on casino win in Atlantic City. Review of Industrial Organization 13 (6): 713-26.

Shonkwiler, J. Scott. 1992. A structural time series model of Nevada gross taxable gaming revenue. Review of Regional Studies 22 (3): 239-49.

Timeline of Terror. 2002. Thinkquest. http://library.thinkquest .org/CR0212088.tertime.htm (accessed January 29, 2005).

Walkup, Carolyn. 2001. Hard landing: Guest traffic tanks after terror attacks. Nation's Restaurant News 35 (41): 1, 65.

Wall Street Journal. 2002. The economy: Las Vegas casinos face union ire--Rich bonuses after layoffs inspire unusual tension, as contracts come to close. May 7, sec. A2.

David Eisendrath is an executive at Harrah's Entertainment (deisendrath@yahoo.com). Bo J. Bernhard, Ph.D., is the director of gaming research at the William E Harrah College of Hotel Administration at the University of Nevada, Las Vegas (bo.bernhard@unlv.edu), where Anthony E Lucas, Ph.D., is an associate professor (afl2@cox.net) and Dennis J. Murphy, Ph.D., is a professor of statistics.

Exhibit 1:
Overview of Selected Terrorist Attacks

Date            Location                      Incident

12/17/73   Rome, Italy/Beirut,   Pan Am Flight 110 bombed prior to
             Lebanon               takeoff by Palestinian terrorists
7/29/79    Madrid, Spain         Two railway stations bombed by
                                   ETA Basque terrorists
6/13/85    Rome, Italy/Beirut,   TWA Flight 847 out of Rome hijacked
             Lebanon/Algiers,      by Hezbollah terrorists
             Algeria
10/7/85    Port Said, Egypt      Achille Lauro cruise ship hijacked
                                   by members of the Palestine
                                   Liberation Organization
4/2/86     Athens, Greece        TWA Flight 840 bombed during
                                   landing, Ezzedine Kassam Unit
                                   of the Arab Revolutionary Cells
                                   claimed responsibility
12/21/88   New York/Lockerbie,   Pan Am Flight 103 bombed, Popular
             Scotland              Front for the Liberation of
                                   Palestine terrorist group claimed
                                   responsibility

Source: D'Amore and Anuza (1986); Timeline of Terror (2002).

Exhibit 3:
Las Vegas Strip Slot Win versus Table & Games Win

Year       Slots (%)   Table & Games (%)

1999         49.2            50.8
2000         49.5            50.5
2001         50.9            49.1
2002         52.4            47.6
2003         53.7            46.3
2004 (a)     54.3            45.7

(a.) Through November 2004 (Nevada Gaming Control
Board 1990-2004).

Exhibit 7:
Box-Ljung Statistics for Autoregressive Integrated Moving
Average (ARIMA) Model (0,1,1) x (1,0,0)

                                   Box-Ljung   Degrees
                        Standard   Statistic     of      Significance
Lag   Autocorrelation    Error       Value     Freedom

1          .006           .074        .006        1          94.0
2          .007           .074        .014        2          99.3
3         (.041)          .074        .321        3          95.6
4         (.030)          .074        .489        4          97.5
5          .055           .073       1.053        5          95.8
6         (.019)          .073       1.118        6          98.1
7          .004           .073       1.121        7          99.3
8          .070           .073       2.031        8          98.0
9          .023           .073       2.134        9          98.9
10        (.083)          .072       3.440       10          96.9
11        (.125)          .072       6.449       11          84.2
12        (.072)          .072       7.450       12          82.6
13        (.180)          .072      13.715       13          39.4
14         .058           .072      14.381       14          42.2
15        (.052)          .071      14.915       15          45.8
16        (.063)          .071      15.709       16          47.3
17         .221           .071      25.413       17           8.6
18         .045           .071      25.821       18          10.4
19        (.058)          .070      26.492       19          11.7
20        (.052)          .070      27.040       20          13.4
21        (.030)          .070      27.226       21          16.4
22        (.001)          .070      27.226       22          20.3
23         .078           .070      28.470       23          19.9
24        (.012)          .069      28.501       24          23.9
25         .034           .069      28.742       25          27.5
26        0.110           .069      31.296       26          21.7
27        (.084)          .069      32.807       27          20.4
28        (.170)          .068      38.981       28           8.1
29         .024           .068      39.109       29          10.0
30        (.112)          .068      41.814       30           7.4
31         .073           .068      42.976       31           7.5
32        (.037)          .068      43.284       32           8.8
33        (.051)          .067      43.854       33           9.8
34         .035           .067      44.129       34          11.4
35        (.052)          .067      44.724       35          12.6
36        (.054)          .067      45.391       36          13.6
37        (.022)          .066      45.503       37          15.9
38        (.018)          .066      45.579       38          18.6
39        (.054)          .066      46.253       39          19.8
40         .100           .066      48.590       40          16.5
41         .046           .065      49.093       41          18.1
42        (.013)          .065      49.131       42          20.9
43         .157           .065      54.965       43          10.4
44        (.067)          .065      56.042       44          10.5
45        (.045)          .064      56.530       45          11.6

Note: Negative numbers appear in parentheses.

Exhibit 9:
Significant Las Vegas Strip Property Openings

                                        Approximate Casino
Date            Property        Rooms   Size (Square Feet)

11/22/89   Mirage               3,039        107,200
06/19/90   Excalibur            4,032        110,000
10/15/93   Luxor                2,526        100,000
10/26/93   Treasure Island      2,900         83,800
12/18/93   MGM Grand            5,005        171,500
06/21/96   Monte Carlo          3,002         90,000
01/03/97   New York--New York   2,023         84,000
10/15/98   Bellagio             3,005        155,000
03/02/99   Mandalay Bay         3,643        135,000
05/03/99   Venetian             3,036        120,000
09/01/99   Paris Las Vegas      2,916         85,000
08/18/00   Aladdin              2,567        100,000

Exhibit 11:
Box-Ljung Statistics for Autoregressive Integrated Moving Average
(ARIMA) Model (0,1,1) x (1,0,0) with Strip Property Openings

                                   Box-Ljung   Degrees
                        Standard   Statistic     of      Significance
Lag   Autocorrelation    Error       Value     Freedom       (%)

1          .038           .074        .258        1          61.2
2          .007           .074        .265        2          87.6
3         (.081)          .074       1.463        3          69.1
4         (.038)          .074       1.724        4          78.6
5          .021           .073       1.806        5          87.5
6         (.024)          .073       1.917        6          92.7
7          .022           .073       2.009        7          95.9
8          .073           .073       3.013        8          93.4
9          .016           .073       3.062        9          96.2
10        (.066)          .072       3.891       10          95.2
11        (.103)          .072       5.917       11          87.9
12        (.065)          .072       6.726       12          87.5
13        (.136)          .072      10.308       13          66.9
14         .065           .072      11.144       14          67.5
15        (.019)          .071      11.218       15          73.7
16        (.039)          .071      11.522       16          77.6
17         .207           .071      20.085       17          27.0
18         .036           .071      20.345       18          31.4
19        (.052)          .070      20.882       19          34.3
20        (.062)          .070      21.654       20          36.0
21        (.004)          .070      21.657       21          41.9
22         .021           .070      21.752       22          47.5
23         .055           .070      22.375       23          49.8
24        (.007)          .069      22.386       24          55.6
25         .009           .069      22.404       25          61.2
26         .109           .069      24.907       26          52.4
27        (.074)          .069      26.063       27          51.5
28        (.139)          .068      30.174       28          35.5
29         .059           .068      30.927       29          36.9
30        (.068)          .068      31.936       30          37.0
31         .073           .068      33.105       31          36.5
32        (.043)          .068      33.503       32          39.4
33        (.054)          .067      34.157       33          41.2
34        (.023)          .067      34.276       34          45.5
35        (.073)          .067      35.469       35          44.6
36        (.073)          .067      36.688       36          43.7
37         .017           .066      36.754       37          48.0
38        (.019)          .066      36.837       38          52.3
39        (.060)          .066      37.673       39          53.0
40         .064           .066      38.619       40          53.2
41         .041           .065      39.005       41          56.0
42        (.015)          .065      39.055       42          60.1
43         .149           .065      44.320       43          41.6
44        (.034)          .065      44.588       44          44.7
45        (.030)          .064      44.810       45          48.0

Note: Negative numbers appear in parentheses.