Method
To analyse the information on expenditure by tourist, we use a two-step approach, in which we first analyse the total daily amount of expenditure, and second the budget shares of various categories in this total. From the point of view of economic theory of consumer behaviour, this approach can be motivated by the assumption that goods consumed during a holiday are separable from other goods in the consumer’s utility function.Footnote 11 This means that we can analyse the expenditure pattern during the consumer’s visit to Amsterdam conditional on the budget that is available for the trip without having to pay attention to the tourists’ consumption behaviour in his or her home country. Our empirical specification of the budget share equations is consistent with the almost ideal demand system of Deaton and Muellbauer (1980). Since we do not have much information about prices (and our observations refer only to 2 years), we estimate no price coefficients, but only Engel curves. We estimate an equation for total daily expenditure per capita, and budget share equations for a number of expenditure categories: lodging, meals/restaurants, transport and parking, shopping, museums visits, theatre visits, cannabis, club/bar visits and other expenditures).
The (economic) importance of the tourism sector for Amsterdam, and other Western European capitals, motivates an interest in the effects of socio-demographic, travel-related and psychological variables on travel expenditures. And especially for policy makers in Amsterdam, our distinction between ‘cannabis tourists’ and ‘cultural tourists’ is potentially of interest. Wang et al. (2006) already pointed out in their survey that research on the attitudes towards and perceptions of destinations and their impacts on spending patterns is a worthwhile topic.
Total daily expenditures
Table 5 presents the results of the regression analyses. Column 1 shows the results of the OLS regression, with total euros spent per person per day as the dependent variable, and in columns 2–9 the share of the total daily amount spent per person (budget share), of each consecutive spending category, is used as the dependent variable. In the present subsection, we discuss the estimation results of the total daily expenditures.
Table 5 OLS pppd expenditures and the nine spending categories by independent variables
The focus of interest is are four trip purpose dummies. We find widely differing coefficients, indicating substantial differences in spending. ‘Other purposes’ has been used as the reference category, and we find that respondents with mixed purposes spend on average more than €30 pppd less.Footnote 12 Those coming to Amsterdam for a special event or an exhibition or for cannabis spend significantly more than this group. It is remarkable that tourists coming mainly for cannabis spend—after controlling for other variables—almost as much per day as those coming for special events or exhibitions. The difference between the daily expenditures of those interested especially in culture, and the reference group is relatively small and not statistically significant. We thus find widely differing amounts spent pppd that are related to travel purposes, confirming Wang and Davidson’s (2010) conjecture.
There are a number of other findings that warrant some discussion. Column 1 of Table 5 indicates that tourists visiting Amsterdam spend less per person per day (pppdFootnote 13), the more days they stay overnight.Footnote 14 The size of the travel party does not have a significant effect on the total amount spent pppd. A similar result is reported by Wang et al. (2006), whereas Thrane and Farstad (2011) and Mok and Iverson (2000) find a negative effect.
Male respondents report to spend more than female respondents, but it is not entirely clear how to interpret this result when respondents do not travel on their own. For instance, most of the expenditure of a couple may be paid by the male, whereas consumption is mostly shared.
Tourists aged 41–50 spend the highest amount of money pppd, followed by those aged 51–60 and 31–40. The oldest and youngest tourists spend the smallest amounts pppd. This is in line with the findings of Thrane and Farstad (2011). The coefficient of the vacation dummy indicates that respondents who are on vacation spend significantly less pppd than respondents who are in Amsterdam for business reasons (the reference category). The difference between the two groups is large because businessmen/women stay in more expensive hotels and travel more often by taxi (AMSTERDAM MARKETING 2012).
The estimated coefficients for the country-of-origin dummies confirm that the respondents with the longest travel distance have the highest expenditures. Respondents from China spend the most money pppd, followed by Russians, while US respondents occupy the third place. Among the Europeans, Scandinavian respondents spend the most money pppd, followed by the Austrians and Brits. However, for the shorter travel distances, the relationship between travel distance and expenditures is less clear. It should be kept in mind that Dutch visitors are the reference category. Their daily expenditure is lower than that of all other categories.
The year dummy for 2011 is significantly positive and indicates substantially higher amounts spent pppd compared with 2006. Note that the expenditure figures have been adjusted for inflation using the cpi figures from the Statistics Netherlands. The financial crisis apparently did not have a negative impact on the expenditure of Amsterdam tourists. And estimated figures for the total number of tourists indicate that in 2011 there were substantially more tourists in 2011 compared with 2006 (Research and Statistics (O&S) Amsterdam 2013).
Budget shares
The equations for the budget shares of the nine categories of spending confirm that there are important differences in the spending patterns of the tourists visiting Amsterdam. We find many significant coefficients and the difference in budget shares can be as large as 7% points (accommodation for those visiting an event or exhibition) of even 8% points (cannabis for those coming especially for that propose). There is only one significant coefficient for trip purpose for the budget shares food, transport and parking, and the rest category. Visitors coming to Amsterdam with mixed purposes spend a larger share of their budget on accommodation and food and less on theatres and museums than the reference group. For those focusing on culture, the differences with the reference group are usually small in % points, although sometimes highly statistically significant. Visitors of events or exhibitions differ most from the reference group in the budget share of accommodation. When cannabis is the main travel purpose, expenditure shares on all other categories except clubbing are negative or not statistically significant.
With respect to trip purposes, we see that respondents travelling to Amsterdam for an event or exhibition—controlling for other variables—allocate most of their daily budget to accommodation, followed by mixed purpose travellers, culture purpose travellers, other purposes, and finally respondents with the cannabis trip purpose. When we focus further on the other shares of spending categories in relation to trip purpose, we see that cannabis tourists allocate their budget only significantly more to buying cannabis and clubbing; the coefficients of the other categories are negative or insignificant. The respondents who come especially for culture allocate relatively more of their daily budget to museums, theatres and accommodation and significantly less to shopping, clubbing and cannabis. This difference in budget allocation shows that these groups are distinct in their spending behaviour and can be targeted differently by tourism marketing and offers.
The Engel curves that we estimate have the logarithm of total daily expenditure (the total spending pppd) as explanatory variable. This is the Working–Leser specification used in the almost ideal model of Deaton and Muellbauer (1980). Estimated coefficients show that a higher pppd budget implies that a larger share of the money is allocated to accommodation, shopping, theatres and clubbing and less to food (confirming Engel’s law), transport and parking, and museums, whereas the share of expenditure on cannabis does not change. The categories with a positive coefficient are luxuries: their budget share increases with the size of daily expenditure. For instance, for accommodation a 50% increase in daily expenditures from €139 (the average in our sample) to €208.50 will lead to an increase of: \( \beta_{{{\text{Ln}}({\text{euro/day}})}} *\ln (\frac{208.5}{139}) = 0.071*\ln (1.5) = 0.029 \) percentage point of budget share. The example shows that the differences in budget shares that are related to travel purposes are large compared to those of the total travel budget.
The log of the number of days stayed in Amsterdam has a negative impact on the budget share for accommodation and a positive one on shopping, while its effect on other categories is more limited. The same remark holds for the size of the travel party.
For the male dummy, we find negative coefficients on shopping and museums and a positive one for cannabis. This is quite suggestive, although for this variable the same caveat holds as for total daily expenditure.
With a higher age, more is spent on accommodation, food and transportFootnote 15 and less on shopping, cannabis and clubbing.
The larger daily expenditure of foreigners is associated with higher budget shares for accommodation and museums and lower ones for food and shopping. It should be recalled that Dutch visitors are the reference.
The coefficients for the 2011 dummy show that the larger daily expenditure in that year was associated with lower budget shares for accommodation (possibly reflecting limited increases in hotel rates due to the crisis) and an equally higher budget share for clubbing.