Terrorist Attacks and International Travel

by Michael Holmes

Terrorism strikes fear into the hearts of individuals. It dominates news coverage and often influences policy following an attack. When attacks occur in developing nations, they can harm international tourism to those nations, presumably striking fear into potential travelers. Terrorists also attack developed nations. If international tourism slows in developing nations, how do tourists respond to attacks in the developed world? Specifically, how did the World Trade Center attacks on September 11, 2001 impact international tourism to the United States?

Studies on terrorism generally claim that a primary goal of terrorism is to create fear. Could fear manifest itself by reducing international travel to the attacked country? International tourism demonstrates a general faith in the relative safety and security of the country a tourist visits. If people believe they risk death or injury by visiting a country, most will travel elsewhere. The September 11^th attacks are unique - using planes as a weapon - in that they could have hurt international travel as a whole, in addition to travel to the United States. I want to quantify the effect the September 11^th attacks had on international tourism to the United States and the rest of the developed world. I hypothesize that terrorist attacks against the United States caused both a small decrease in international travel overall and a substantial decrease in travel to the United States. 

I use the differences in differences method, which compares international tourist travel to the United States with other countries before and after September 11^th_. I compare data on tourism (International Arrivals) to the United States with four similar countries as controls: Australia, Japan, Italy, and France. [1] I chose Australia and Japan because they are similar to the United States in that most international tourists must travel by plane. If September 11^th impacted air travel generally, it should impact all three countries. These countries in particular will allow us to observe the effect the attacks had specific to the targeted country.

Italy and France are slightly different since many international tourists arrive by other methods of transportation, but they are still good comparisons. Like the United States, France and Italy have major museums, well-visited tourist attractions, and a comparable amount of tourist traffic each year. Most importantly, the trends in tourism for these four nations were similar to the United States, even if the overall totals varied (See Figure 1). [2]

In order to make up for some of the differences in tourism totals, we must control for other factors (See Table 1). I control for the previous year’s Real GDP Per Capita because international travel generally requires advanced planning and payment.[3] A poor financial year could negatively influence travel abroad and is particularly relevant since the dot-com bubble burst during this period. I also include variables that measure the effect specific countries have on international tourism.[4] These help control for geographic, historical, and country-specific features that enhance or hinder tourism. The country controls and the Real GDP Per Capita control help account for differences in overall tourism and increase the validity of our measurements.[5]

To perform the differences in differences measure, I create a variable (Post) to mark the date of the attack.[6] I use the year 2002 to start the Post-9/11 period and justify it for a couple of reasons. First, my data is yearly and it is wiser to start in 2002 to measure September 11^th effects since the event occurred more than halfway through 2001. Choosing 2001 as the treatment year risks biasing the estimate, as there are over eight months of other explanatory variables unaccounted for. Second, many schedule international trips in advance, meaning that the true impact of the attacks might not occur until four or five months after the event. In addition to the Post variable, I include a Pre variable to mark the years prior to 2002, a Treat variable (The United States), and a Control, which represents the remaining countries.

I run the regression through 2006, which provides five years of data after the attack. I use 2006 as a cutoff because I believe the detrimental effect on the tourism industry was temporal and I want to avoid the inclusion of too many additional historical events that could influence international tourism such as the banking crisis or the weakening of the dollar.

International tourism to the United States decreased by 9.8% after 2001, but international tourism to the control countries increased by 10.1%.[7] The United States had more international tourism overall than the control nations both before and after 9/11, but while other nations experienced growth, tourism declined in the United States. Compared to international tourism in the other countries, international tourism to the United States fell 18.1%.

These results partially confirm my hypothesis. There was not a broad negative impact on international tourism, but there was a negative impact on tourism to the United States. While the attacks appear the likely cause, alternative theories are still possible. Increased security and anti-terror procedures in the United States after 2001 made flying more of a hassle, possibly discouraging potential travelers. While that theory does not take public fear into account, the increased security is still a reaction to the terrorist attacks. Other explanations could involve economics. However, these also tend to favor the terrorist attacks as the most likely cause for the tourist decline. The Dollar fell against the Euro from late 2001 through 2003. This means that vacations to the United States were cheaper for Europeans during this period, but tourism still declined.[8]  While there are some other explanations, the September 11, 2001 terrorist attack was at least partially responsible for the decline in international tourism in the United States from 2002-2006. Terrorist attacks can cause damage to the tourism industry in even the strongest of nations. Since an attack in one country in an effective means of causing fear on an international scale, terrorist organizations can be expected to continue to target both developed and developing nations.

Appendix:

The percentage values are created by interpreting the log values of the tourist arrivals. The 18.1% decrease in international tourism to the US compared to the differences in other countries is the percentage change on the actual number of tourists, not the log values.

You may notice that the first value under the Difference column in Table 3 is negative, where as the corresponding coefficient in Table 2 is positive. It is negative in Table because it is measuring the difference going from the United States Pre 9/11 to Post 9/11 (a 0 to 1 shift). As a coefficient in the regression table, it measures the increase in Y going from 0 to 1, and therefore has a positive value. 

---

[1] All countries had a high rating on the Human Development Index around September 11, 2001. I chose a Human Development Index score of .80 as my floor for selecting control countries.

[2] In order to evaluate the common trends assumption for these countries, I performed a correlation test in addition to the graphs shown above. The correlation coefficients between the United States and the following countries before 2002 are as follows: Australia=.8228; Japan=.8129; Italy=.8101; France=.7748. Correlations coefficients from 1995 to 2006 are still positively correlated, but lose strength. This is possibly a result of the hypothesized relationship.

[3] Performed a varsoc test and received results recommending a one- or four-year lag. The one-year lag made the most sense theoretically, as a previous year’s income is probably more indicative of the next year’s spending capabilities.

[4] These are dummy variables based on Country Code.

[5] Performed a VIF test for check for multicolinearity, all variables are below a 10.

[6] Post is a 1 if the Year is greater than or equal to 2002. It is a 0 if it is 2001 or earlier. 

[7] -.1033 significant at 10% level and .0956 significant at 1% level. See Appendix for full Regression tables.

[8] Unfortunately the data does not say where tourists came from, so it is possible that Europe increased their US tourism and the rest of the world declined. However, this scenario also means that it was now more expensive for non-Europeans to travel to Europe, yet the European countries in the control experienced an increase.