*2.3. Studies on Tourism Expenditure Using Other Models*

In recent years, in order to better understand tourists' expenditure behavior, some researchers have employed new modeling frameworks to perform in-depth analyses. D'Urso et al. (2020) propose the fuzzy double-hurdle model, which combines the doublehurdle model with fuzzy set theory to take into account the effect of satisfaction on tourists' expenditure behavior. The new model allows the researchers to handle the imprecision of both collected information (i.e., levels of satisfaction) and the kind of measurement used (i.e., a Likert-type scale). Pellegrini et al. (2021) investigated tourists' expenditure behavior by implementing a framework that jointly adopts the stochastic frontier (SF) regression and multiple discrete–continuous extreme value (MDCEV) models. This framework allows the researchers to not only identify the maximum level of spending that the individual is willing to incur but also to assess two interrelated decisions: whether to allocate a budget for a specific expenditure category as well as the amount to be spent on that chosen category. Besides, a conditional quantile regression model has been applied in identifying leisure tourism expenditure patterns (e.g., Alfarhan et al. 2022).

In addition, other explanatory factors that may influence tourists' decision-making have been considered using various analytical techniques. Park et al. (2020) applies different estimation procedures, namely, ordinary least squares (OLS), two-stage least squares (2SLS), the Heckit model, and quantile regression (QR) to perform an analysis of the determinant factors in relation to total expenses. The role of information sources in predicting travel spending behaviors represents new possibilities for analyzing the determinants of expenditure by using QR. Chulaphan and Barahona (2021) investigated the determinants of tourist expenditure per capita in Thailand by utilizing an autoregressive distributed lag model (ARDL) and using panel-estimated generalized least square (EGLS). Such knowledge is essential for tourist authorities to develop profitable and sustainable

tourism projects in destinations whose natural resources have been affected by profitseeking tourism. ism projects in destinations whose natural resources have been affected by profit-seeking tourism.

terminant factors in relation to total expenses. The role of information sources in predicting travel spending behaviors represents new possibilities for analyzing the determinants of expenditure by using QR. Chulaphan and Barahona (2021) investigated the determinants of tourist expenditure per capita in Thailand by utilizing an autoregressive distributed lag model (ARDL) and using panel-estimated generalized least square (EGLS). Such knowledge is essential for tourist authorities to develop profitable and sustainable tour-

*Economies* **2022**, *10*, x FOR PEER REVIEW 4 of 22

#### *2.4. Proposed Research Framework 2.4. Proposed Research Framework*

According to the two-stage decision model, the decision on the intention to use tourist accommodation and that of accommodation expenditure constitute the consumer behavior of tourist accommodation. Based on a summary of the previous literature on tourism expenditure (e.g., Dardis et al. 1981, 1994; Cai 1999; Nicolau and Màs 2005; Sun et al. 2015) and by considering the implementation of vacation policy, the variables influencing the intention to use and actual expenditure on tourist accommodation can be classified into six categories, namely, the economic factor, social stratum, geographical location, family life cycle, tourism behavior, and vacation policy. In this study, it is assumed that the economic factor influences the expenditure on tourist accommodation but does not influence the intention to use accommodation. This is mainly because if the same explanatory variable is included in the two sets of decision equations, it may be impossible to correctly identify the model's parameters (Newman et al. 2001). Therefore, it is necessary to add certain exclusion restrictions (Jones 1992; Newman et al. 2001; Aristei et al. 2008) to facilitate the estimation of the parameters in the model equations. In terms of the empirical application, it is usually assumed that the participation equation is a function of noneconomic factors; thus, the economic factor can be excluded from this equation (Newman et al. 2001; Aristei et al. 2008). The research framework of this study is presented in Figure 1. The research hypotheses are presented as follows. According to the two-stage decision model, the decision on the intention to use tourist accommodation and that of accommodation expenditure constitute the consumer behavior of tourist accommodation. Based on a summary of the previous literature on tourism expenditure (e.g., Dardis et al. 1981; Dardis et al. 1994; Cai 1999; Nicolau and Màs 2005; Sun et al. 2015) and by considering the implementation of vacation policy, the variables influencing the intention to use and actual expenditure on tourist accommodation can be classified into six categories, namely, the economic factor, social stratum, geographical location, family life cycle, tourism behavior, and vacation policy. In this study, it is assumed that the economic factor influences the expenditure on tourist accommodation but does not influence the intention to use accommodation. This is mainly because if the same explanatory variable is included in the two sets of decision equations, it may be impossible to correctly identify the model's parameters (Newman et al. 2001). Therefore, it is necessary to add certain exclusion restrictions (Jones 1992; Newman et al. 2001; Aristei et al. 2008) to facilitate the estimation of the parameters in the model equations. In terms of the empirical application, it is usually assumed that the participation equation is a function of noneconomic factors; thus, the economic factor can be excluded from this equation (Newman et al. 2001; Aristei et al. 2008). The research framework of this study is presented in Figure 1. The research hypotheses are presented as follows.

**Figure 1.** The research framework for the two-stage decision model of the intention to use and consumption expenditure on tourist accommodation. **Figure 1.** The research framework for the two-stage decision model of the intention to use and consumption expenditure on tourist accommodation.

#### 2.4.1. Participation Decision 2.4.1. Participation Decision

According to Nicolau and Màs (2005), Jang and Ham (2009), Alegre et al. (2013), and Bernini and Cracolici (2015), there is a positive link between the tourism participation decision and an individual's education level. Indeed, higher educational levels may provide training and preparation for some types of recreational activities (Dardis et al. 1981) and also easier access to information and knowledge (Cai 1998). Such information and knowledge are likely to increase the desire to discover new destinations and enjoy new experiences (Bernini and Cracolici 2015). Furthermore, individuals with a high level of education are more likely to reach adequate job positions and a higher level of income, which could be spent on non-basic needs like tourism. Occupation status was found to be a significant

social discriminating factor in tourism participation (Bernini and Cracolici 2015). Thus, we propose the following hypothesis:

**Hypothesis H1a.** *The social stratum has a significant impact on the intention to use tourist accommodation.*

In Jang and Ham's (2009) study, the variables of age and marital status were found to be significant for the travel decisions of elderly seniors. The research findings of Alegre et al. (2013) indicated that a positive effect was detected for tourism participation in the case of the presence of children in the household. Bernini and Cracolici (2015) found that the tourism participation decision was affected by cohort effects: the oldest cohorts were more inclined to participate in tourism than the youngest ones. The empirical results of Sun et al. (2015) indicated that the family travel intention varies at different stages of the household life cycle. Therefore, we hypothesize the following:

**Hypothesis H1b.** *The family life cycle has a significant impact on the intention to use tourist accommodation.*

By referring to Cai (1998), Nicolau and Màs (2005), Jang and Ham's (2009), and Bernini and Cracolici (2015), the empirical analysis has emphasized the role of population location and consequently the attributes of the tourists' region of residence. These studies have found that geographical variables are significant to the tourism participation decision. In a wider sense, the residential area takes in both territorial differences in tourism resources and socio-economic differences among residents' living conditions. Therefore:

**Hypothesis H1c.** *The residential area has a significant impact on the intention to use tourist accommodation.*

Four variables have been selected to represent tourism behavior, including days of the trip, travel season, travel date, and favorite activity during the trip. Li et al. (2021) revealed that tourists' behaviors in selecting travel seasons and the associated trip duration were influenced by a few factors and the correlation between these two tourism decisions was conditional upon the covariates. Dellaert et al. (1998) argued that tourists may be restricted by school holidays when choosing the period in which to travel. Indeed, time factors, including time convenience, were the most often cited reasons for not participating in recreational tourism (McGuire 1984). The finding of Wu et al. (2011) indicated that time constraints reduced the number of long trips, the number of short trips, and, to a greater extent, travel intention.

Tourists expect to recover more completely during a vacation by removing themselves from daily settings and actively engaging in various restful activities. Laybourn (2004) stated that the decision-making of festival participants may be associated with personal factors, such as lifestyle. Nicolau and Màs (2005) concluded that a greater propensity to go on holiday was associated with a favorable opinion of going on holiday. Both lifestyle and tourists' favorable opinions may reflect on their engagement in a certain activity which implies the benefits they seek (Moscardo et al. 1996). Those who seek more benefits from leisure and recreational activities may tend to lay emphasis on the high quality of travel and the use of accommodation. Thus, we hypothesize the following:
