Impact of exchange rates on air travel A. Edward Day Department ofEconomics, Universityof Central Florida, Orlando, FL 32816, U.S.A.
The impact on the travel and tourism industry of changes in exchange rates has not obtained the same level of interest as that enjoyed by manufacturing. However, U.S. tourism competes with the world economy for travelers in the same way that the automobile manufacturers must. Increases in the value of the dollar make foreign travel more attractive to U.S. citizens and U.S. travel less attractive to foreigners. This paper provides some estimates of the impact on air travel of changes in the U.S. rate of exchange for 22 countries. The findings are that exchange rates have been important in nearly all cases, with most demand elasticities ranging between 1 .O and I .5. Estimated demand elasticities can be used to estimate the impact on net air travel of percentage changes in exchange rates.
Key words:
exchange rates
air travel
impact of changes Given the number of individuals whose employment depends on a strong tourism industry and the number of locales in the U.S. that depend to a significant degree on the performance of this industry, it is surprising that more work has not been done in this area. This is especially true for the recent period when fluctuation in the value of the dollar has increased.
Spending by foreign tourists in the U.S. grew from $6.2 billion in 1977 to $12.9 billion in 1981, an increase in real terms of approximately 68%. HOWever, these expenditures fell by 12% between 1981 and 1983. From 1977 to 1981 U.S. tourists’ spending overseas increased from $7.5 billion to $11.5 billion, an increase in real terms of about 13%. From 1981 to 1983 these expenditures had increased by an additional 22%. Clearly in the more recent period, foreign tourists decreased their spending in the U.S. while American tourists spent more abroad. The causes and consequences of this change in the flow of tourism expenditures are important to the U.S. tourism industry.
In a recent study, Zalatan and Burdge (1980) identified several factors important in determining decisions by Americans to travel abroad. These included real U.S. incomes, the share of leisure expenditures in income, tourism expenditures as a proportion of leisure expenditures and the price of tourism as measured by relative price levels in the U.S. and 21 foreign countries. Missing from this list of important factors is the rate of exchange.
Changes in the value of the dollar in international markets are commonly believed to have an important effect on the relative demand for U.S. goods and services. In general, however, these arguments are concerned with the major industries, such as steel and automobiles.
Artus (1972) studied the effect on tourism expenditures of changes in relative price levels, and exchange rates for the countries of western Europe, Canada and the U.S. His study is broader than the present one in that Artus attempts to model all intercountry flows. Artus uses per capita expenditure on travel rather than numbers of travelers.
The purpose of the present study is to investigate the impact of changes in exchange rates on international travel to and from the U.S. The focus of the research is on the air traveler because data on other modes of travel is not readily available.
Both of the above studies are concerned with global rather than regional issues of tourism flows. Thus, Zalatan and Burdge (1980) address questions of the factors affecting the growth of tourism worldwide. Artus’ concern is with total intercountry flows of tourism expenditures.
The tourist industry which directly employs 6 million workers is an important sector of the economy that is affected by changes in the value of the dollar. For example, if the value of the dollar increases it has the dual effect not only of discouraging foreign visitors to U.S. attractions but also providing an incentive for Americans to visit abroad rather than vacation in the U.S. Int.J. Hospitality
Management Printed in Great Britain
A major benefit of the Artus study is its theoretical development. However, the database that he
Vol. 5 No. 3 pp. 115-l 19,1986
02784319iE6 $3.00 + 0.00 Pergamon Journals Ltd 115
116
A. Edward Day
is not rich enough to support the theoretical model. Therefore, most of the equations are estimated with no more than 10 degrees of freedom. For this reason, most of the empirical work may be of questionable value.
NETTRAV
The present research is more narrowly focused on net travel to the U.S. and how those travel decisions are affected by changing currency values. Additionally, using a previously unexploited database, demand estimates for more countries are possible than in the Artus study. Since quarterly data is available for all countries studies, sufficient observations exist to validate the observations made.
CPIR
uses
The principal finding of the present study is that changes in the value of the dollar in the international markets have had an important impact on the decisions of U.S. citizens to travel abroad and on the decisions of foreign visitors to visit the U.S. In addition, U.S. price levels have played a major role in this decision-making. In the next section of the paper the estimating equation will be described. Then, the data sources and definitions of variables will be explained. Next, the estimates of the equation’s parameters for each of the countries will be provided. The final section of the paper gives an interpretation of the results, and possible extensions of the research. The demand equations estimated in this study are straightforward applications of demand theory and follow the work of Artus to a large degree. The principal difference between Artus’ work and this one is that his primary concern was with the levels of travel expenditure demand between countries, i.e. an attempt to model to total flows of spending, while the task of this work is to estimate the net flow of travel only to the U.S. market. The demand for foreign travel by U.S. citizens is a function of relative domestic and foreign prices, U.S. income levels, and the dollar exchange rate. The demand for travel to the U.S. by foreign nationals is a function of the same set of variables looked at from their perspective. in this study, the ratio of U.S. departures to foreign arrivals is used as a measure of net demand for travel to the U.S. This measure was chosen because the data indicate that the variance of the logarithms of both departures and arrivals are more stable than that of the nontransformed series. The estimating equation takes the general form: NETTRAV where:
= F(IPR, C&R, E&R, NETI)
IPR
EXR NET1
= natural logarithm of the ratio of U.S. nationals’ departures to foreign nationals arriving in the U.S. = natural logarithm of the ratio of foreign income to U.S. income. = natural logarithm of the ratios of foreign price level to U.S. price level. = natural logarithm of the exchange rate. = NETTRAV lagged one period.
The expected signs of the partial derivatives with respect to each of the independent variables are indicated in the equation above. The variable NET1 was required in some of the equations to correct for problems of autocorrelation of the residuals. This method was suggested by &anger and Newbold (1975). In the following section estimates are provided for the net demand for travel to the U.S. This demand function is estimated for 22 countries and regions. The results are reported for two groups. The first group is countries for which complete data were readily available on prices, income and exchange rates. Only exchange rates and U.S. price and income data were available for the second, larger set of countries, The results for the second group are no less impressive than for the first group. Quarterly data on arrivals and departures at U.S. ports of entry were obtained from the Immigration and Naturalization Service I-94 reports, IIa and IId covering the period from 1976: I to 1984 : IV. Monthly exchange rates, industrial production and consumer price indexes were taken from the CITIBASE economic database. Quarterly averages were formed from the monthly data. Preliminary estimation indicated that the demand equations were not linear in the nontransformed data. Therefore, the variables were transformed into logarithms. A benefit of estimation in this form is that the coefficients of the model represent elasticities of demand with respect to the respective variables. Table 1 reports the results of OLS regressions for the first group of countries. Table 2 reports regressions for the second. The dependent variable is the difference of the logarithm of U.S. departures and the logarithm of foreign arrivals. Independent variables for the first group are natural logarithms of the ratios of industrial production (a proxy for national income), ratios of U.S. and foreign CPIs, and exchange rates. Independent variables for the second group of countries are modified because of a lack of readily
117
lmpactofexchange rateson airtravel
Table 1. Regressions on group 1 countries Country
Const.
Japan
-4.136 (2.177) 0.275 (0.251) -0.377 (2.314) -8.406 (2.855) 6.258 (3.099)
France Germany Italy U.K.
IPR
CPIR
EXR 0.615 (1.837) -0.585 (0.510) 1.133 (6.780) 1.454 (3.098) -1.044 (3.146)
-
-
3.827 (0.819) 0.778 (3.0219) -2.067 (2.936) -1.566 (1.59)
-2.394 (0.882) -1.273 (2.021) -2.418 (3.290) -0.175 (0.203)
NET1 0.319 (1.870) -
a*
DW*
F
0.19
2.40
4.88
0.15
2.28
1.86
-
0.64
1.66
19.48
-
0.45
1.49
8.73
0.56
1.81
11.66
0.334 (2.000)
* Durbin-Watson statistic.
available
rates for the U.K. is reversed because the variables are expressed as U.S. cents per pound.) The Durbin-Watson statistic is within the acceptable region. The F-ratio is significant at the 5% level of confidence or better for all equations except France. Adjusted R2values for all but France and Japan seem acceptable.
data. Only U.S. industrial production and CPI are used along with the exchange rate. Proxies for relative income and price were not consistently available over the time period for most of these countries. There are few surprises in the results for the first group of countries. All the signs for exchange rates are as expected and with the exception of France, are statistically significant. (The sign on exchange
Equations for Japan, Germany, Italy and the U.K. indicate that the elasticity of demand with
Table 2. Regressions on group 2 countries Country Africa Argentina Belgium Brazil Columbia Denmark Ireland Chile Caribbean Mideast MidAmerica SouthKorea Sweden(B) Phillipines(B) Peru(B) Netherlands(B) Venezuela(B)
Const. -3.881 1.339 8.924 (3.294) -0.603 (0.712) 13.815 (3.605) 8.982 (6.789) 1.447 (0.743) 11.509 (2.533) 1.849 (0.808) 2.121 (1.138) -6.038 (2.457) 3.280 (2.367) -5.210 (2.409) 14.771 (4.111) 1.632 (1.458) 11.810 (5.170) 1.571 (1.577) -0.148 (0.111)
EXR 6.447 (2.690) 0.221 (4.760) 1.200 (6.622) 0.316 (3.788) 1.955 (7.984) 1.157 (2.729) -0.888 (2.013) 0.396 (2.092) -0.202 (1.192) 8.874 (5.193) 0.344 (4.227) 0.430 (2.556) 1.879 (2.722) 1.934 (6.816) 0.515 (6.048) 1.024 (3.065) 0.649 (1.631)
CPIR -0.704 (2.414) -2.120 (3.788) -0.714 (3.648) -2.826 (3.871) -3.163 (7.962) -0.736 (1.592) -1.192 (2.502) -0.739 (1.424) -0.329 (0.960) -0.799 (4.503) -0.773 (2.651) 0.396 (0.531) -3.372 (4.164) -1.089 (3.939) -2.787 (5.541) -0.444 (2.084) -0.236 (0.741)
NET1 0.090 (0.501) 0.230 (1.400) 0.143 (0.851) 0.385 (2.166)
0.208 (1.400) 0.506 (3.063) -0.253 (1.522) 0.223 (1.260) 0.539 (3.273)
I?*
DW
F
0.44
1.93
10.02
0.43
1.98
12.29
0.57
1.99
22.06
0.61
1.77
18.90
0.67
1.88
32.96
0.20
1.95
4.16
0.17
1.80
3.38
0.15
1.92
2.84
0.18
1.88
3.634
0.63
1.98
28.03
0.60
1.83
18.18
0.62
2.27
19.90
0.36
1.65
9.09
0.71
2.21
28.75
0.52
1.70
18.45
0.47
1.82
11.00
0.51
1.91
12.80
A. Edward Day
118
Table 3. Total net travel
to U.S.
Const.
EXR
CPIR
NET1
R2
DW
- 1.407 (2.126)
0.546 (2.490)
-0.200 (1.376)
0.644 (5.020)
0.73
2.13
respect to the exchange rate for travel between these countries and the U.S. is between 0.62 and 1.50, approximately. This means that a 1% increase in the value of the dollar in these international market will have the effect of reducing the flow of travelers to U.S. destinations between OS and 1.5%. This reduction. is from increased U.S. citizens traveling abroad and foreign nationals’ decreased travel to the U.S. In terms of the average net travel to the U.S. over the sample period, a 1% increase in the value of the dollar would have resulted in a net annual decrease in travelers available to the U.S. tourist industry of approximately 2.8 million. For the second group of countries, the coefficient for the exchange rate variable in all equations has the correct sign and is statistically significant. The surprise in these equations arises from the sign on the U.S. price variable, CPIR. Since price indices for all the countries were not available, only the U.S. price level was used. This variable entered most equations with a presumably incorrect sign but statistically significant. The exceptions are the Caribbean, Korea and Venezuela. An interpretation of this result is that the U.S. CPI is an index of world price levels. When the CPI in the U.S. is rising, prices in the other countries and regions in this study were increasing also, but at a faster rate. Thus, one observes the negative effective on the ratio of U.S. departures to foreign arrivals when the U.S. price level goes up. Significant coefficients of elasticity of demand for net travel with respect to the exchange rate ranges from a high of 6.4 for Africa to 0.22 for Argentina. On average, these elasticities appear to be lower than for the group 1 countries. This could be the result of mispecifying the equations, and reestimation with data on foreign prices might increase the size of these coefficients. Estimates for both the group 1 and group 2 countries confirm the importance to international travel decisions of relative prices and exchange rates. This information can be useful to those in the tourist industry in their attempts to plan for the vacation season. The final estimate produced for this study is for all net travel between the U.S. and all countries except Canada. In this estimate, the trade weighted index of the dollar is used as the exchange rate
F 31.58
along with the U.S. CPI as independent variables. Table 3 reports the result of this regression. These few variables explain almost three-quarters of the variance in total net travel to the U.S. over the 48 quarters of the sample. The exchange rate is significant and has the expected sign. The U.S. CPI has an insignificant effect on the net demand for travel. Lagged net travel (NETl) has an important effect on current travel and is necessary in the equation to correct for autocorrelation of the residuals. The elasticities of demand for air travel to the U.S. reveal the importance of international exchange rates to the performance of the tourism industry. The extent to which a number of regions of the U.S. economy depend on this industry means that these regions can be importantly affected by changes in relative currency values. To the extent that changes in exchange rates are the result of national economic policy choices, the industry depends to a significant degree on stability in these national policy choices. One of the more practical uses to which these results can be put is to estimate demand scenarios during the tourist season. To do so requires an estimate of expected percentage changes in the value of the dollar, relative price levels (CPIs) and relative income levels between the U.S. and other countries. Pptting these values into the equation then will yield the expected changes in the foreign market. Information from this exercise can then be used to direct marketing efforts to those countries most sensitive to exchange rate fluctuations. An extension of the present work would be the construction of a bridge from the effect of exchange rates on national travel and tourism flows to their effect on local tourism demand. This extension would incorporate differences in the cost of travel between competing locales. These cost differences would be both local cost factors and differences in cost of travel from ports of entry to the competing sites. This extension is presently under way.
- Support for this study was provided by a grant from the Division of Sponsored Research, University of Central Florida.
Acknowledgement
Impact of exchange
rates on air travel
References Artus, J. R. (1972) An econometric analysis of international travel flows. Znternational Monetary Fund Staff Papers 19, 579613. @anger, C. W. J. and P. Newbold (1975) The time series
119 approach to econometric
model building. New Methods in
Business Cvcle Research: Proceedins
from a Conference,
Sims, C. A. (ed.), pp. 10-13. Fed&al Reserve Rank of Minnesota, Minneapolis. Zalatan, A. and Burdge R. (1980) The future of international tourism: some neglected factors. Joumalof TravelResearch, Spring, 15-19.