The Demand For Package Tours:

The Demand For Package Tours: A Mode Choice Model PAULINE J. SHELDON AND JAMES MAK A model is presented to explain a traveler’s choice of vacation mo...
Author: Martha Wiggins
13 downloads 0 Views 514KB Size
The Demand For Package Tours: A Mode Choice Model PAULINE J. SHELDON AND JAMES MAK

A model is presented to explain a traveler’s choice of vacation mode. The possible modes considered are independent travel, travel on an inclusive package tour, and travel on a basic package tour. The model is tested using logit analysis and survey data on travel to the Hawaiian Islands. The results indicate that purchasers ofpackage tours are likely to be elderly, be intent on visiting several destinations, contain few people in the party, intend to make short visits, and be first-time visitors to the destination.

Package or prepaid tours are an important component of the travel business. A package tour is a combination of the many components of a vacation such as transportation, accommodation, sightseeing, and meals which are sold to customers at a single price. Consumers may not know the prices of the individual components because they purchase the entire package either from a travel agent, who is a retailer of vacation products and not usually responsible for creating tour packages, or from a tour operator, who creates the package and publishes, and then distributes the brochures. However, in most cases, the component parts can be purchased separately through a travel agent or directly from the suppliers. Presently, package tours account for one-third of all U.S. travel abroad. A survey conducted by Touche Ross and Co. (1975) of over 1,000 U. S. tour purchasers found that 8% of them would not travel if package tours were unavailable, and another 42% would significantly change their travel plans. (The Touche Ross study makes a distinction between escorted, hosted, and package tours, but we include all three in the concept of package tours here.) Despite the importance of package tours in the travel industry, we are aware of only two significant studies that have been done on their demand. Askari ( 1971 ) employed sales data from a single (unnamed) large U. S. tour operator to estimate the number of people from each state purchasing package tours. He found that demand for package tours was highly sensitive to per capita state income, the price ofthe tour per day, and the number of attractions per day. He did not conduct an analysis of why some people chose to travel on package tours while others traveled independently. The second study, the Touche Ross survey ( 1975,p. 18), solicited responses from tour purchasers on the reasons why they selected tours and yielded the following results:

27

Other reasons

100%

Convenience, the most frequently cited advantage of package tours, was especially important to older travelers. This finding is not surprising since travel is strenuous and older travelers might find the physical requirements of independent travel (e.g., baggage handling) too demanding. Tour economy was the second most frequently cited reason for purchasing a tour, and among those who purchased tours that were not escorted or hosted, tour economy was the overriding factor (43%) for tour selection. The Touche Ross study, however, did not present a unified model explaining the travelers’ choice of independent travel vs. travel on package tours. The study cited the benefits of taking tours, but not the disadvantages, and the data set included only those who took a tour and not those who traveled independently. In this article we develop a binary choice model which explains a traveler’s choice of independent travel vis-a-vis travel on package tours. We test the model using survey data collected by Hawaii Visitors Bureau (HVB) on U. S. resident travel to Hawaii. About 40% of the U. S. visitors to Hawaii

purchase package tours. MODEL In our model, the relevant decisionmaking agent is the visitor party, not the individual visitor. (We may assume that the visitor party equals unity with no loss of generality.) We assume that a visitor party has decided to visit a particular destination. The remaining decision is the model oftravel, i.e. travel on a tour or travel independently. A model for this consumer decision can be developed as follows. A visitor (i) will choose to travel on a certain vacation mode (j) (package tour or independent travel) with some This probability is determined by certain probability consumer attributes, certain trip attributes, and certain modal attributes and can be expressed as follows:

P~.

Paulme J. Sheldon Is Assistant Professor of Tourism at the School of Travel Industry Management, University of Hawaii. James Mak is Professor of Economics at the Umversity of Hawaii. The authors would like to thank Evelyn Richardson, Research Director of Hawaii Visitors Bureau, for the data used in this paper.

Pij

=

fi [Xi, Tj,Mj]

(1)

where Xi is a vector of consumer attributes for consumer i, Tj is a vector of trip attributes and Mj is a vector of modal attributes for mode j. 13

The survey conducted by Touche Ross and Co. found convenience to be the most frequently cited advantage of package tours. This suggests that package tour travel is more attractive to certain consumer types. Older travelers may especially value the convenience when making their mode choice. Single women may prefer to travel on a package tour for safety and security reasons. First-time visitors to a destination generally have higher information needs than repeat visitors so that the potential benefits from purchasing a package tour are greater for first-time visitors. Also, since popular attractions are offered in many tours a repeat visitor who has experienced them in previous visits might not wish to buy them again. Hence, we posit that repeat visitors are less

likely to purchase package tours. For some individuals, however, package tours may represent limitations. Wealthy travelers may demand luxurious services that are not available in package tours. The consumer attributes affecting the vacation mode choice can therefore be represented as follows:

Modal Attributes The prices of the various vacation modes certainly will affect the consumer’s choice of mode. Many travelers perceive package tour travel to be cheaper than independent travel. Of the respondents in the Touche Ross survey ( p. 21 ), 56% believed that package tours were cheaper than similar travel arrangements booked separately and only 11 % believed them to be more expensive. On the other hand, the cost of travel on package tours rises with the size of the tours are priced on a per person person pays the same price. By contrast, visitors who travel independently may enjoy substantial

traveling party. Package basis, and each

economies of scale in the purchase of some kinds of vacation goods. For example, hotels often permit children to stay free with adults, whereas on a package tour, children are required to pay separately. Indeed in some cases, young children are not permitted on a tour. Therefore, one attribute of tour packages is that they are less attractive to larger party sizes and especially to parties traveling with children. Since many tours have fixed arrival and departure dates, visitors who desire schedule flexibility will not choose package tours but rather will travel independently. However, many tour operators are now providing more flexibility in components and arrival and departure dates to accommodate these visitors. The modal attributes affecting the vacation mode choices can be represented as follows:

Trip Attributes Three trip attributes may affect the choice of vacation mode. They are the length of stay, the number of destinations visited on the trip, and whether the trip is to a foreign destination. Visitors expecting to stay at a destination for only a brief period might find it more advantageous to purchase package tours. Since time is scarce, the costs of searching and buying vacation goods are correspondingly high. Buying a package tour will minimize these costs to the consumer and enable the visitor to &dquo;see more and do more.&dquo; In traveling to unfamiliar or foreign destinations or traveling to multiple destinations, visitors who purchase a tour can reduce uncertainty and lower the search and transactions costs in gathering information and arranging a trip. The traveler who travels independently can also reduce these costs by employing the services of a travel agent( see Mak and Moncur 1980), but the costs cannot be reduced below the costs of arranging package travel since arranging independent travel through a travel agent would still require many more separate personal transactions than buying a single package. This is especially true for multiple destination trips. The trip attributes affecting the vacation mode choice can therefore be represented as follows:

14

The probability of a visitor choosing a package tour ( PKG=1 ) over independent travel ( PKG=0) can then be written

as:

(42%) purchased tours. The mean price of all the package tours was $1,597. We computed the equivalent retail costs of packages if the components had been purchased separately (Hawaii Visitors Bureau estimated the retail prices of most components using published retail price lists from suppliers). Since tours also included round-trip airfares to Hawaii, we used the 1980 edition of the Official Airline Guide (North American edition) to estimate the airfare components of the package tours. We used the lowest available excursion fare (Y class) from the visitor’s home city. The mean value of the equivalent retail cost of the average package tour to Hawaii would have been $1,886, so that the average discount (PDIFF) on package tours was $289 ($1,886-$1,597), or 15.3% (=$289/$1,886 x 100). The 15.3% package tour discount is probably a lower bound estimate since we made no attempt to estimate the equivalent retail values of tour guide or escort services if such services were included in the packages. the

where the

signs above the variables indicate the expected relationship. Equation (5) can be estimated using either logit or probit analysis (see Pindyck and Rubinfeld 1981 ). We choose to use logit analysis since that has been most commonly used in modal choice modeling and has become the standard. It also facilitates the calculation of the elasticities of the explanatory variables.

Supply of Vacation Modes It should be noted that supply factors may also affect the traveler’s choice of vacation mode. Independent travel is always available to travelers either through travel agencies or by direct contact with the suppliers. Package tours, however, vary in their availability with the destination and time of year under consideration. Popular destinations are likely to have more tour operators servicing them, so the consumer has more variety of tour packages to choose from. Prices are also likely to be lower because of the strong competition. The supply of tour package options will also vary with the season of the year, with more packages available in the high season. Tour operators may also create discount packages to increase business when suppliers’ occupancy rates are especially low. (For a full discussion of tour operators see Sheldon 1986).

DATA

EMPIRICAL MODEL In empirically applying the model developed above to the destination of the Hawaiian Islands, a few modifications are made. As all respondents were from the U. S. mainland or Canada and did not experience any currency or language problems, the &dquo;foreign destination&dquo; variable was dropped. The empirical model does not include a variable for young children, since an extensive review of tour brochures revealed no prohibitions against children joining tours to Hawaii. Finally, the &dquo; repeat visitor&dquo; dummy variable was replaced by the more comprehensive &dquo;trip number&dquo; variable representing the total number of visits to the Hawaiian Islands. After making these modifications, the empirical model for visitors to the Hawaiian Islands is as follows:

The model is tested using data collected by Hawaii Visitors Bureau in a year-long survey in 1980 of westbound (mostly U. S.) visitors to the Hawaiian Islands. All westbound visitor parties to Hawaii were asked to voluntarily complete a short survey attached to the Agricultural Declaration Form prior to leaving their airplane. The survey requested information on the sociodemographic characteristics of the visitor party, home residence, and hotel accommodations while in Hawaii. From those responding (about 55%), a sample of 7,194 visitor parties were generated for a follow-up survey. Daily logs were delivered to all the selected parties at their hotels within one or two days of their arrival. One log was used per traveling party. Recipients were asked to indicate whether they came to Hawaii on a package tour and, if so, the name of the tour, the price paid, and the components of the tour. In addition, all visitors, including independent travelers, were asked to record their daily expenditures while in Hawaii. The logs were mailed back to Hawaii after visitors returned home and a small souvenir was sent to those who responded. A total of 1,749 logs were returned for a return rate of 24.3%. Among the respondents, 59% were traveling independently and 41 1 % came on package tours. An analysis of the respondents indicated that younger visitor parties were under-represented in the sample; hence the proportion of independent travelers was probably understated. For our analysis we removed from the sample all business travelers (35), visitor parties staying with friends or relatives (66), all foreign visitors (105) who were predominantly intransit, and responses with either missing key data or bad data (251). We were left with a total of 1,292 usable responses. Among them, 750 (58%) traveled independently and 542 15

independent travel ~

senior citizens

~

multi-destinational travel

likely to purchase package tours generally longer visits

Less ~

EMPIRICAL RESULTS We estimated equation (6) using logit analysis. The independent variable in the regression equation is simply the logarithm of the odds that package tours will be selected

Prob( Package)

i. e., log ~i.e., 1- Prob(Package)1°

The model

was

estimated using

maximum likelihood estimation. All variables were tested for multicollinearity by inspecting the correlation matrices, and none was found. The overall model explains the data well ( R2=.41 ) and was found to be significant at the 1 % level using the chi square test as shown in Table 1. The individual coefficients were tested using the t-test, which is appropriate because the large sample size causes the parameter estimates to be asymptotically normally distributed. (See Pindyck and Rubenfeld 1981, p. 311). All the coefficients, except for SINGLE WOMAN, have the predicted signs and are significantly different from zero at the 5% level. Table 1 summarizes the logit analysis results and shows the elasticities for each variable. TABLE 1

RESULTS OF LOGIT ANALYSIS FOR THE VACATION MODE CHOICEa

~

the rich

~

larger visitor parties

~

repeat visitors

The elasticities for each of the independent variables show that changes in a traveler’s wealth and age have the most effect on mode choice. The party size, the number of destinations visited, and the number of times the traveler has been to Hawaii come next in significance, with price differences having the least impact. To provide a simple illustration of these results, consider two hypothetical visitor parties, A and B, with the following characteristics:

Party A ~

party head is elderly

~ ~ ~

intends to visit two islands two people in the party first visit to Hawaii

Party B ~ ~ ~ ~

party head is

not

elderly

intends to visit only one island four people in the party second visit to Hawaii

We further assume that neither party is rich, both intend to stay for seven days, and the price difference between package tour travel and independent travel for identical trip attributess (PDIFF) is $250. Using our logit estimates, we determine that visitor party A has a.85 probability of visiting Hawaii on a package tour compared to .46 probability for visitor party B. In other words, A is likely to purchase a tour, but B is likely to travel independently. Further, for a given visitor party, the probability of purchasing a tour diminishes as PDIFF declines. For visitor party A, if the discount for package tour travel is eliminated, the probability that A will purchase a tour falls from .85 to .68; for B, the probability falls from .46 to .24. In sum, visitors are somewhat responsive to changes in the

package tour prices. BASIC VS. INCLUSIVE TOURS

point, we have only considered the choice package tour travel and independent travel. However, tours vary considerably in their &dquo;inclusivity.&dquo; The appropriate modal choice model must also consider the different tour options available to consumers. The Touche Ross study occasionally differentiated between escorted, hosted, and package tours. Askari ( 1971 ) ignored these distinctions by including an independent variable in his demand equation that explicitly controlled for the &dquo;number of attractions per day.&dquo; Among all package tours, the simplest is the basic tour which typically includes transportation, baggage handling, and accommodation only. Inclusive tours also offer some sightseeing and entertainment at the destination; all-inclusive tours include meals and sometimes are escorted. Among the 542 visitor parties in our data set that visited Hawaii on Up

to this

between

Likelihood ratio test = 465.421 with 8 d.f. R2 = .41. Elasticities are calculated using AP, = a, (P,(1-P,) ) with 0.42 (the mean value) for P,. (see Pindyck and Rubinfeld 1981, pp. 299-300.) * Significant at the .05 level. a

b

In brief, our empirical results on the visitor choice of independent travel versus package tour travel can be summarized as

follows:

More likely to purchase package tours . the greater the package tour price discount 16

over

package tours, 227 (or 42%) purchased inclusive tours. Although specific evidence is lacking, there is consensus among Hawaii travel experts that the proportion of Hawaii visitors traveling on inclusive tours has declined significantly during the last decade. In theory, it is possible to construct a multiple-mode choice model( see Pindyck and Rubinfeld 1981, pp. 301-310) in which visitors choose from among an array of travel options that include independent travel, basic tours, and inclusive tours. Unfortunately, we were unable to construct appropriate price variables for such a model. Our second-best approach was to estimate a binary choice model in which package tour travelers choose between basic and inclusive tours. The logit equation to be estimated is as follows:

dependent variable is the logarithm of the odds of a package tour traveler choosing an inclusive tour instead of a basic tour. Table 2 summarizes the results of the logit analysis. The

TABLE 2

RESULTS OF THE DECISION TO PURCHASE AN INCLUSIVE TOURa

tions

less likely to purchase inclusive tours. By contrast, traveling to several islands are more likely to purchase inclusive tours. The elasticities show that the addition of one additional destination increases the probability most of all that the traveler will choose an inclusive tour. Between 1974 and 1984, the average length of stay of westbound(i.e., mostly U.S.) visitors to Hawaii has remained virtually unchanged; hence it cannot explain the decline in the proportion of visitors purchasing inclusive tours. The probable explanation is the rise in the percentage of repeat visitors from 37.3% to 47.3% and the decline in the proportion of the visitors desiring to visit several islands. are

tour visitors

CONCLUSION We have developed a unified model that explains a traveler’schoice of independent travel vis-a-vis travel on package tours. The model was tested using logit analysis and survey data collected by Hawaii Visitors Bureau on U. S. resident travel to Hawaii. In contrast to earlier studies, our analysis includes data on both package tour and independent travelers. The principal limitation of our study is that our empirical (but not our theoretical) results are not directly applicable to other destinations which are differentially served by package and non-package tours. However, the theoretical model, which does include supply factors, is more applicable to other destinations. Our empirical results are quite robust. They indicate that visitor parties who are elderly, intend to visit several destinations, contain few people in the party, intend to make short visits, and are on their first trip to Hawaii are more likely to purchase package tours than to travel independently. The study also found that travelers are responsive to the price savings afforded by package tour travel. Our results also suggest that, for Hawaii, the continued rise in the percentage of repeat and single destination visitors will diminish the future demand for package-tour travel. The decrease in demand will be particularly evident among the inclusive tours. The trend in Hawaii’s visitor industry clearly will be directed toward independent travel and to basic tours, a fact which tour wholesalers must recognize to remain

competitive. REFERENCES a

Likelihood ratio test

*

Significant at the

=

17.815 with 6 d.f. R 2= .03.

.10 level.l.

Askari, Hossein (1971), "Demand for Package Tours," Journal of Transport Economics and Policy (January), 40-51. Hawaii Visitors Bureau (1974-1984), Annual Research Report, 19741984, Honolulu.

These results are not as robust as the earlier ones. The model is, however, significant at the 5% level as shown by the likelihood ratio test. Only NUMDEST is significant at the 5% level. However, the coefficients of TRIP NUMBER and LSTAY have the correct signs and are both significant at the 10% level. The results indicate that repeat tour visitors are less likely to purchase inclusive tours compared to first-time tour visitors. Similarly, tour visitors who plan lengthy vaca-

Mak, James and James E. T. Moncur (1980), "The Demand for Travel Agents," Journal of Transport Economics and Policy (May), 221-231.

Official Airline Guide (1980), North American edition. Pindyck, Robert S. and Daniel L. Rubinfeld (1981),

Econometric

Models and Economic Forecasts, 2d. ed., New York: McGraw Hill. Sheldon, Pauline J. (1986),"The Tour Operator Industry: An Analysis," Annals of Tourism Research 13 (July), 349-365. Touche Ross and Company (1975), Tour Wholesaler Industry Study, New York: Touche Ross.

17