The Impact of the Business Cycle on Service Providers

Abstract

For service providers, it is essential to understand how their business is affected by the macroeconomy. This is especially pressing for the tourism sector, the world’s largest export service, because the number of incoming visitors is likely to be strongly determined by the business cycles in the countries of origin. Utilizing state-of-the-art business-cycle metrics, we derive novel insights on the relationship between international tourism and the business cycle. We find an excess sensitivity of the sector to economic cycles based on a multidecade data set of international visitors to New Zealand coming from multiple counties and with various visitor purposes. However, we find no asymmetries in the speed of adjustment across contractions and expansions, suggesting a quicker recovery than many other (nonservice) sectors. Moreover, a higher cyclical volatility results in higher growth in the long run. A robustness check for two more destination countries (Australia and Japan) yields comparable insights. The results underscore the need to closely monitor the cyclical sensitivity and long-term growth prospects of the various visitor streams into the country, in order to (i) better tailor the accommodations and services to these streams and (ii) exploit diversification opportunities to reduce the overall cyclical volatility.

Keywords

business cycle export services co-movement elasticity long-term growth cyclical sensitivity

Introduction

The recent Global Financial Crisis (GFC) has reminded many businesses of their sensitivity to macroeconomic factors and has inspired a variety of studies on the impact of business-cycle fluctuations (see Tellis and Tellis 2009, for a review). This research has led to insights on the impact of economic contractions and expansions on a variety of marketing metrics, such as industry or brand sales (Deleersnyder et al. 2004; van Heerde et al. 2013), market share (Steenkamp and Fang 2011), private-label share (Lamey et al. 2007, 2012), and price and/or advertising elasticity (Gordon, Goldfarb, and Li 2013; van Heerde et al. 2013). However, these studies have primarily focused on either consumer durables (Deleersnyder et al. 2004) or consumer packaged goods (Gordon, Goldfarb, and Li 2013; Lamey et al. 2007, 2012; van Heerde et al. 2013). The cyclical sensitivity of services, in contrast, has received much less research attention in the marketing literature.¹

The Cyclical Sensitivity of Services

This lack of attention is surprising because the service industry contributes significantly to a country’s gross domestic product (GDP) and employment levels. For example, the service industry in the United States accounts for more than 80% of U.S. nonfarm employment (Rust and Chung 2006), and close to 75% of the country’s GDP (Kumar et al. 2014). Moreover, the service industry differs on key attributes from the physical goods studied in extant business-cycle research, which makes a study on the cyclical sensitivity of the service industry especially relevant.

Foremost, because of the inseparability of its production and consumption, and its inherent perishability (Zeithaml, Parasuraman, and Berry 1985), a service cannot be stored, neither by manufacturers nor by consumers (Filardo 1997). When a downturn leads to a drop in demand for services (e.g., fewer beauty treatments or less frequent lawn mows), service providers are unable to temporarily produce for inventory. Goods manufacturers, in contrast, can smooth their production and employment by creating additional stock during periods of low demand (Lebow and Sichel 1995). Moreover, the production of services often requires a relatively high labor component (Cuadrado-Roura and V.-Abarca 2001) compared to the production of goods. As a consequence, a drop in the use of services during a recession tends to have a relatively strong impact on the amount of work available for frontline personnel. To make matters worse, the skills of an organization’s most knowledgeable frontline employees, once laid off, may be hard to replace by newcomers when demand picks up again (Lovelock 1997), which could affect service firms’ growth prospects during subsequent expansion periods.

Based on these observations, marketing researchers have posited that the service sector can be expected to be more cyclically sensitive than the goods sector. Kumar et al. (2014, p. 673, italics added), for example, argue that “service industries are especially vulnerable to the state of the macroeconomic environment.” A similar assessment is made in Lovelock (1997, p. 14) who posits that “service firms are particularly threatened” in recessionary times. Still, there is little empirical evidence for this excess sensitivity.² Lovelock, for example, only offered conceptual guidelines on how service firms could best prepare for future recessions. Kumar et al. (2014) based their conclusion on the 2008–2011 contraction, which some have argued is not representative for the service sector’s cyclical dependence in general (De Nardi, French, and Benson 2012; Petev, Pistaferri, and Eksten 2012). To avoid inferences driven by the idiosyncrasies of a single downturn, we study an extended time span of several decades that covers multiple contraction and expansion periods.

Important differences are likely to exist in the cyclical sensitivity between different types of services (Zeithaml, Parasuraman, and Berry 1985). Some services are more discretionary than others (Lovelock 1997), making it easier to postpone purchases. Similarly, the consumption of certain luxury services is more susceptible to social-visibility considerations (Kamakura and Du 2012), and also the income elasticity differs widely across service types (Falvey and Gemmell 1996). As such, we do not consider the combined (aggregated across all possible service types) employment or expenditure series for the service industry, as sometimes done in labor-economic studies (e.g., Filardo 1995, 1997) or macroeconomic studies (e.g., Cook 1999), but rather focus on one of the most important export services, international tourism. We study not only the level of cyclical sensitivity of this crucial service sector but also the speed of adjustments to, respectively, contraction and expansion periods, and the long-run growth implications of the sector’s sensitivity. We also explore the heterogeneity in its cyclical sensitivity across different countries of origin and visitor purposes. These refinements enrich our understanding on how service sectors can ride the economic tide.

The Cyclical Sensitivity of International Tourism

This article studies international tourism, looking at the number of international visitors for different visitor purposes (holiday, business, and Visiting Friends and Relatives [VFR]). We adopt the classification of the United Nations World Tourism Organization (UNWTO), which states “A visitor is a traveler taking a trip to a main destination outside his/her usual environment, for less than a year, for any main purpose (business, leisure or other personal purpose) other than to be employed by a resident entity in the country or place visited. These trips taken by visitors qualify as tourism trips” (United Nations [UN] 2008, ¶ 2.9).³

International tourism is a crucial revenue source to many transport services (such as airlines, cruise ships, and taxicabs), hospitality services (such as hotels and resorts), and entertainment venues (such as amusement parks, casinos, and shopping malls). Globally, tourism is the fourth largest export sector (after fuels, chemicals, and food) and the largest service export sector, accounting for as much as 30% of the world’s exports of commercial services (UNWTO 2012, p. 3). Tourism’s contribution to GDP amounts to around 5%, and up to 10% for several countries, as illustrated in Table 1.

Table 1.

Impact of Tourism on National Economies Across the World (2012 Share).

Region	Country	Travel and Tourism Contribution to Service Industry (in %)^a	Travel and Tourism Contribution to Employment (in %)^{a, b}	Travel and Tourism Contribution to Export (in %)^a	Travel and Tourism Contribution to National GDP (in %)^a
Africa	South Africa	16.9	10.3	10.0	9.7
Americas	Brazil	13.3	8.2	2.6	9.1
	Canada	6.2	5.2	3.2	4.3
	United States	10.7	10.0	8.0	8.5
Asia	China	20.6	8.3	2.3	9.2
	India	10.0	7.7	4.1	6.5
	Japan	9.2	6.9	1.6	6.7
	Thailand	37.6	12.7	12.1	16.6
Europe	France	12.2	10.8	8.0	9.7
	Germany	6.6	4.7	2.5	4.5
	Spain	21.5	15.5	13.8	15.2
	United Kingdom	8.7	7.5	5.1	6.8
Oceania	Australia	14.2	11.9	6.8	9.8
	New Zealand	21.4	19.1	13.5	14.9
World		14.5	8.7	5.3	9.2

Note. CIA = Central Intelligence Agency; GDP = gross domestic product.

^a Source. World Travel and Tourism Council (http://www.wttc.org/research/economic-data-search-tool/). ^b Source. CIA World Factbook (https://www.cia.gov/library/publications/the-world-factbook/fields/2012.html).

However, unlike some services (Filardo 1997), international tourism is often sensitive to international economic conditions (although less so if the purpose is to visit family and friends, as discussed in more detail later). Also, international tourism is less essential than some other services, and characterized by high social visibility, which (as discussed below) also influences the extent of cyclical sensitivity.

Despite a prolonged growth in visitor numbers, there is a mounting concern among policy makers about the implications of recessions. In highly tourism-dependent regions, any reduction in the number of incoming visitors will add to already stringent fiscal and balance-of-payments constraints. During the first three quarters of 2009 (shortly after the Lehman Brothers bankruptcy), the number of international visitors to the world’s top 50 tourist countries dropped by more than 8% relative to the same period in 2008 (BusinessWeek 2009). For countries such as Greece and Spain, visitor flows have been estimated to drop by as much as 30% in 2009, contributing significantly to these countries’ bleak economic situation (European Network for Accessible Tourism 2009). Given such numbers, it is not surprising that business analysts speculate about the impact of economic contractions and expansions on the international-tourism sector. The contraction side tends to drive the headlines, with titles such as “The Recession Doesn’t Spare Travel and Tourism” ( The Wall Street Journal 2009) or “Global Travel Takes a Dive” (BusinessWeek 2009).

In spite of an abundance of popular press articles and sector white papers, little systematic knowledge is available on how the business cycle affects international visitor streams. Most of the available evidence is either anecdotal, or idiosyncratic to a single crisis, making it hard to reach any rigorous, generalizable conclusions. Frechtling (1982), for example, focused exclusively on the impact of the 1981–1982 U.S. recession, while Kumar et al. (2014) studied the 2008–2011 financial crisis. While academic studies have recognized the importance of economic factors when explaining/forecasting the demand for tourism (Balaguer and Cantavella-Jordá 2012; Mazanec 1986; Song et al. 2010; Song, Witt, and Li 2009), the impact of business-cycle fluctuations has hardly been studied.

In this study, we systematically investigate the relationship between the cyclical state of the economy and the international tourism using data spanning multiple expansion and contraction periods. What complicates matters, however, is that international visitors come from a variety of countries of origin, which may well be in very different stages of their business cycle. For example, Japan experienced a deep economic downturn in 1993–1995, while Australia experienced a strong upswing in that same period. Similarly, the GFC of 2008–2009 caused severe troughs in the GDP (per capita) of countries such as Japan, the United States, and the United Kingdom, but much less so for China and Australia. As such, there is not just one business cycle to be considered but a different one for each country of origin.

Moreover, not all visitor streams may be equally sensitive to the changing economic situations in their home country. This sensitivity may differ not only across countries (e.g., in function of the distance to be traveled to reach a destination country) but also in function of the purpose of visiting (e.g., business visitors vs. VFR). If so, the composition of a country’s visitor base can vary substantially depending on the changing economic status in different regions of the world. An understanding of this composition, and of its evolution, is crucial to better tailor local service offerings to current and expected visitor needs (Coviello, Winklhofer, and Hamilton 2006). For example, a higher proportion of visitors from a certain region may call for more service providers speaking the visitors’ native language (Holmqvist and Grönroos 2012) and tailoring menus in restaurants (Moran, Harris, and Moran 2007). Similarly, business and holiday visitors tend to have different needs in terms of transportation, accommodation, and entertainment, as do visitors with different cultural backgrounds.

Knowing how economic contractions and expansions affect the total number of incoming visitors and its composition across various countries of origin and visitor purposes should help the sector to better ride the global economic tides. Against this background, we address the following four novel research questions: (i) Does international tourism exhibit a more pronounced cyclical sensitivity than the economy as a whole? (ii) Does the cyclical sensitivity in international visitor streams differ across (a) visitor purposes and (b) countries of origin? (iii) Does the speed of adjustment in international tourism differ across economic contractions and expansions, that is, is it asymmetric? and (iv) Do business-cycle-induced swings in visitor numbers hurt or benefit the long-run growth prospects of international tourism?

Theoretical Background and Research Questions

Extent of Cyclical Sensitivity

Given that economic contractions reduce people’s wealth (Mehra 2001), put more pressure on companies’ balance sheets (Dobbs, Karakolev, and Maligne 2002), and negatively affect consumer and business confidence (Allenby, Jen, and Leone 1996), they are expected to reduce international tourism. As potential visitors’ disposable income decreases, they may postpone or cancel their trip, or there could be a substitution effect from international toward domestic tourism (Dolnicar et al. 2008). In times of economic expansions, opposite patterns tend to emerge.

Similarly, many firms feel the need to cut costs and preserve cash flows in economically difficult times (Webber, Buccellato, and White 2010), and therefore impose more stringent travel policies and/or substitute business trips by cheaper alternatives. In a recent study by the Business Travel Coalition, half of the 200 companies that were surveyed stated that they were actively seeking alternatives to air travel, such as videoconferencing (The Telegraph 2008).

Importantly, given that tourism is often a discretionary (nonessential) component in consumers’ overall spending (Webber, Buccellato, and White 2010), it may well be affected excessively by economic downturns (Katona 1975). Moreover, international tourism has high social-cultural visibility. A consumer’s desire to spend in such positional categories is reduced disproportionally during economic contractions, because there is less need to still spend (as much) on international tourism to preserve their social standing if others also reduce their expenditures (Kamakura and Du 2012). As for companies, herding behavior may facilitate cost cutting already implemented by competitors during economic downturns (Deleersnyder et al. 2009; Steenkamp and Fang 2011). As such, both economic (discretionary nature) and sociological (public visibility and herding behavior) factors lead us to the first research question:

Research Question 1: Does international tourism exhibit a more pronounced cyclical sensitivity than the economy as a whole?

Differences Across Visitor Purpose and Countries

Cyclical considerations may be less of an issue when VFR than when visiting a destination for holiday or business purposes. Because of the personal dimension involved, the trip becomes more essential (less discretionary), making people less likely to postpone or cancel their trip because of unfavorable economic conditions. In addition, compared to international holiday tourism where multiple destinations can meet consumers’ needs, the destination is likely to be less substitutable for VFR. While leisure-based tourism can be seen as a luxury that is not necessary, the mind-set for VFR is different, which could make it less susceptible to economic conditions (Backer 2012).

Variations in cyclical volatility cannot only be expected in terms of visitor purpose but also across countries. Not all countries of origin are in a similar stage of economic development, and business cycles are not perfectly aligned or synchronized across countries (Cerqueira 2013). Also the sensitivity of consumers’ and managers’ decisions to economic up- and downswings can differ, depending (among other things) on their national culture (Deleersnyder et al. 2009) or the distance to be traveled to the target country (longer distance travel may be more likely to be canceled during a downturn). These observations lead to the second research question:

Research Question 2: Does the cyclical sensitivity in international visitor streams differ across (a) visitor purposes and (b) countries of origin?

Asymmetry in Speed of Adjustment

Many industries evolve asymmetrically across economic up- and downswings. Deleersnyder et al. (2004), for example, documented how consumers quickly reduce their expenditures on consumer durables during economic downturns, while being much slower to increase them again once the economy recovers. Lamey et al. (2007) found a similar pattern in the substitution between national brands and private labels. Such asymmetric behavior, often referred to as steepness asymmetry, can be attributed to various factors.

First, during contractions, consumers’ willingness to buy decreases sharply, as they have a strong incentive to delay their discretionary spending, or to switch to cheaper alternatives. Consumer confidence in the economy, an important determinant of consumers’ spending (Katona 1975; Ou et al. 2014), has been argued to be lost easily but to take longer to restore (Nooteboom, Berger, and Noorderhaven 1997). Moreover, consumers are likely to keep on delaying their nonessential purchases when the economy starts to recover, to take full advantage of their anticipated income or wealth increase (Caballero 1993). Hence, while downward adjustments can occur swiftly, upward adjustments can take considerably longer.

A similar steepness asymmetry has been documented among firms. While managers are quick to save costs in an effort to improve their bottom line when the economy goes sour (Dobbs, Karakolev, and Maligne 2002), they are reluctant to increase that spending again (e.g., on business travel) when economic conditions improve (Deleersnyder et al. 2009). It is unclear, given their differential mind-set, whether the same arguments apply to VFR visitors. This discussion leads us to the third research question:

Research Question 3: Does the speed of adjustment in international tourism differ across economic contractions and expansions, that is, is it asymmetric?

Long-Run Growth Implications

In light of the recent GFC, and the general tendency to focus more on negative than on positive information (Kramer 2002), it is not surprising to see downturns receiving more attention in recent business communications. To the sector, it is key to determine what recessions imply for its long-run growth prospects. Within the economic literature at large, opposing theories exist on the link between the volatility in a country’s economic indicators and their long-run growth. Arguments and empirical evidence can be presented for a negative, a positive, and a null effect of volatility in an economic series (e.g., GDP or visitor numbers) on long-run growth (in GDP or visitor numbers).

A negative relationship happens if a reduction in the number of incoming visitors leads to layoffs of skilled labor in the sector (Lovelock 1997). In addition, businesses may find it difficult to obtain credit to finance their investments in bad economic times (Döpke 2004), which could lead to less private-sector investments.

Other theories postulate a positive relationship. Economic downturns can help to “clean up” the sector, by forcing out the less-fit vintages, leaving more room for the more productive ones. This argument finds its roots in the theory of creative destruction (Schumpeter 1939), which posits that innovation is the force that sustains long-term economic growth. In addition, governments often become more inclined, in an effort to save jobs, to launch stimulus packages and invest public funds in infrastructure development—which can increase the long-run attractiveness of a destination.

Finally, there are also economic theories that postulate that business-cycle volatility and long-run growth are phenomena that are more or less independent, either by inherent assumption (as in neo-Keynesian theories) or because the positive and negative influences cancel one another out empirically (Döpke 2004). This debate inspires the fourth research question:

Research Question 4: Do business-cycle-induced swings in visitor numbers hurt or benefit the long-run growth prospects of international tourism?

Modeling Framework

Our research framework consists of the following steps: First, we apply the Hodrick and Prescott (1997) filter (henceforth, HP filter) to separate the cyclical component in the time series of interest (international visitor numbers and GDP/capita series) from their overall trend or long-run evolution. We subsequently derive key metrics to describe the series’ sensitivity to the business cycle: (i) their cyclical volatility, (ii) their co-movement elasticity, and (iii) their extent of cyclical asymmetry. In addition, we quantify the long-run growth rate in each series. Next, we meta-analytically combine these insights across countries and visitor purposes into a set of empirical regularities and assess their heterogeneity. Finally, we explore the link between the cyclical properties of the series and its long-run growth rate.

Step 1: Extraction of the Business-Cycle Component

Only part of the overtime variation in international tourism can be attributed to business-cycle fluctuations. In line with the prior literature (see, e.g., Cook 1999; Lamey et al. 2007), we use the HP filter to extract those fluctuations that occur at business-cycle periodicities. The filter decomposes a time series (y_t ) into two components: (i) a (long-run) trend component that varies smoothly over time ( $y_{t}^{LR}$ ) and (ii) a cyclical component ( $y_{t}^{c}$ ). To ensure the comparability across the different countries, we analyze the log of y_t , so that the units of $y_{t}^{c}$ (multiplied by 100) represent percentage deviations from the series’ growth path (Stock and Watson 1999).

Step 2: Summarizing the Cyclical Behavior of the Individual Series

To quantify the extent of the cyclical variations in the different visitor series, we derive two summary statistics: (i) the cyclical variability (volatility) and (ii) the degree of cyclical co-movement with the general economic activity in their country of origin. The cyclical variability is operationalized as the standard deviation of the cyclical series, $σ_{i j} ({Visitors}_{i j, t}^{c})$ , with ${Visitors}_{i j, t}^{c}$ the cyclical component in the number of international visitors coming with visitor purpose i (holiday, business, and VFR) from country j, and t the time period. Given that the cyclical components represent percentage deviations, these standard deviations are comparable across series (Deleersnyder et al. 2004, p. 358). To benchmark this variability to the variability for the overall economic activity, we normalize it to the standard deviation of the cyclical component of the corresponding GDP series, $σ_{j} ({GDP}_{j, t}^{c})$ , leading to $σ_{i j} ({Visitors}_{i j, t}^{c}) / σ_{j} ({GDP}_{j, t}^{c})$ .

However, the variability measure does not consider to what extent ups and downs are synchronized with the overall economic cycle. To do so, we derive for each visitor series a co-movement elasticity by regressing the cyclical component in the number of visitor arrivals $({Visitors}_{i j, t}^{c})$ on the cyclical component of country j’s GDP per capita ${GDP}_{j, t}^{c}$ :

{Visitors}_{i j, t}^{c} = β_{i j} \times {GDP}_{j, t}^{c} + ∊_{i j, t} with t = 1, \dots, T .

Since both ${Visitors}_{i j, t}^{c}$ and ${GDP}_{j, t}^{c}$ represent relative deviations from the underlying long-term trends, β _ij can be interpreted as an elasticity. Also, since both series have zero means due to filtering, it is not necessary to include an intercept in Equation 1. As the business-cycle filter may induce serial correlation in the data (Engle 1974), we allow for an autoregressive error term when needed, based on the Durbin-Watson statistic.⁴ The sign of the co-movement elasticity (β _ij in Equation 1) indicates whether international visitor flows move in the same (positive β _ij ) or opposite (negative β _ij ) direction to the economic activity in the country of origin. Its magnitude reflects the extent to which cyclical fluctuations in the source country’s economy are attenuated or amplified in its international visitor patterns. If it is larger than one in magnitude, there is excess sensitivity to the business cycle.

We also assess to what extent the patterns are symmetric across the contraction and expansion stages of the business cycle. While international tourism may drop quickly as a country’s economic situation deteriorates, potential visitors may be reluctant to immediately spend their increased discretionary income on international tourism once their economy starts to recover. In line with Sichel (1993), we test for such steepness asymmetry through the third moment of the filtered series’ first difference:

{ST}_{i j} ({Δ Visitors}_{i j, t}^{c}) = \frac{[T^{- 1} \sum_{t = 1}^{T} {({Δ Visitors}_{i j, t}^{c} - \bar{{Δ Visitors}_{i j, \cdot}^{c}})}^{3}]}{σ_{i j} {({Δ Visitors}_{i j, t}^{c})}^{3}},

with ${Δ Visitors}_{i j, t}^{c}$ the first difference of the cyclical component of the tourist series at hand, $\bar{{Δ Visitors}_{i j, \cdot}^{c}}$ and $σ_{i j} ({Δ Visitors}_{i j, t}^{c})$ the corresponding mean and standard deviation, and T the sample size.⁵

Step 3: Combining the Evidence Meta-Analytically

To increase the power of the inferences, we meta-analytically combine the evidence across the different countries and visitor purposes through Rosenthal’s method of added Z’s (Lamey et al. 2007, 2012; Rosenthal 1991; van Heerde et al. 2013). For the co-movement elasticity, we meta-analytically test both for the procyclical nature of the cyclical fluctuations (i.e., are the values greater than zero) and for the excess sensitivity postulated in Research Question 1 (i.e., are the values greater than one). For the steepness statistic (Research Question 3), we test for combined evidence of negative skewness across the different countries. Following van Heerde et al. (2013), for the effect size of the respective summary statistics, we report their weighted average across countries, with the inverse of each value’s standard error, normalized to one, as weight. The reported effect sizes can therefore be interpreted as a reliability-weighted mean, where values with a higher reliability (lower standard error) receive a higher weight.

Step 4: Quantifying the Long-Run Implications

To quantify the long-run implications of cyclical fluctuations in visitor numbers, we calculate the long-run growth in each tourist series as the average (over time) of ${Δ Visitors}_{i j, t}^{LR}$ , which is the first difference of the trend component.⁶ We regress this average on the metrics derived in Step 2: (i) the relative cyclical volatility in the tourist series $[σ_{i j} ({Visitors}_{i j, t}^{c}) / σ_{j} ({GDP}_{j, t}^{c})]$ , (ii) the co-movement elasticity $[{\hat{β}}_{i j}]$ , and (iii) its steepness statistic $[{ST}_{i j} ({Δ Visitors}_{i j, t}^{c})]$ , while controlling for the average long-run growth rate in the country’s GDP $(\bar{{Δ GDP}_{j, \cdot}^{LR}})$ and the average long-run growth rate in the country’s population $\bar{{(Δ Pop}_{j, \cdot}^{LR})}$ :

\bar{{Δ Visitors}_{i j, \cdot}^{L R}} = γ_{0 i} + γ_{1 i} \times \frac{σ_{i j} ({Visitors}_{i j, t}^{c})}{σ_{j} ({GDP}_{t}^{c})} + γ_{2 i} \times {\hat{β}}_{i j} + γ_{3 i} \times {ST}_{i j} ({Δ Visitors}_{i j, t}^{c}) + γ_{4 i} \times \bar{{Δ GDP}_{j, \cdot}^{LR}} + γ_{5 i} \times \bar{{Δ Pop}_{j, \cdot}^{LR}} + ∊_{i j} .

We allow for separate regression coefficients per visitor purpose, guided by pooling tests, which explains the subscript i for the γ’s. To address the estimated nature of the dependent variable and to obtain efficient estimates, we use weighted least squares (WLS), with the inverse of the dependent variable’s standard error as weight. As some of the independent variables are estimated quantities too, we obtain unbiased estimates for the parameters’ standard errors through a bootstrap algorithm, using 250 Monte Carlo simulations on each of 1,000 random resamples with replacement (for details, see Nijs, Srinivasan, and Pauwels 2007, appendix A2 or the technical appendix of Lamey et al. 2012).

Empirical Evidence

Primary Research Setting

We address the research questions in the context of the international visitor streams to New Zealand. We study 30 different countries of origin, over more than three decades with multiple expansion and contraction periods, and across three different visitor purposes.⁷ We use annual visitor numbers from 30 key countries of origin across the period 1980–2013 (34 years) from Statistics New Zealand (http://www.stats.govt.nz/infoshare/),⁸ accounting for 93% of the international visitors into New Zealand over this period. Because of New Zealand’s remote island nature, all travel is by air or by boat, resulting in excellent records of all incoming visitors. Moreover, tourism is the country’s second largest export industry (Statistics New Zealand 2012), making an adequate assessment of the tourism’s cyclical sensitivity of obvious importance.

Later in this article, we discuss results for Australia and Japan. Following prior research with a few cross-sectional cases (e.g., Ailawadi, Pauwels, and Steenkamp 2008; Guyt and Gijsbrechts 2014), we prefer an in-depth discussion for one case (New Zealand), and use the other cases (Australia and Japan) as robustness checks. The reason we pick New Zealand as the focal country is that (i) New Zealand is the only country for which we have tourism marketing spend, which we will use in one of the other robustness checks and (ii) tourism is, compared to the size of its economy, relatively most important for New Zealand (Table 1).

The countries of origin for tourists visiting New Zealand include 11 European countries (Austria, Denmark, France, Germany, Italy, Ireland, Spain, Sweden, Switzerland, the Netherlands, and the United Kingdom), 2 North American countries (Canada and the United States), 1 South-American country (Brazil), 11 countries from Asia (China, Hong Kong, India, Indonesia, Japan, Malaysia, Singapore, South Korea, the Philippines, Thailand, and Taiwan), 4 countries (Australia, Fiji, Samoa, and Tonga) from Oceania, and, finally, South Africa. Apart from this geographic diversity, the countries also represent a different mix in terms of their stage of economic development. Figure 1 illustrates that there is considerable variation in the levels and in the overtime evolution of the visitor numbers originating from these countries.

Figure 1.

Overtime evolution in the number of holiday and business visitors to New Zealand originating from top visiting countries and growing visiting countries.

To represent the economic activity in each country, we use data on its real GDP per capita (obtained from http://www.imf.org/external/data.htm). Since business cycles typically last between 1.5 and 8 years (Christiano and Fitzgerald 1998), the length of the time series (34 years) ensures that multiple cycles are covered, including the U.S. savings and loan crises of the 1980s and 1990s, the 1990 Scandinavian banking crisis, the 1997 Asian financial crisis, the 2008–2010 Irish banking crisis, and the recent GFC. Combined with the geographic diversity in the countries of origin, this time span allows us to move beyond the idiosyncrasies of any specific crisis and subsequent recovery, and to identify empirical regularities on the nature of macroeconomic business cycles’ impact on international tourism.

The business cycles in the different countries are not fully synchronized. The correlations between the various cycles range from −0.56 (between The Philippines and Tonga) to 0.88 (United States and Canada), with an average (median) value of 0.30 (0.31), well below unity.

Information on the population in the various source countries is obtained from the U.N. statistical database (http://esa.un.org/unpd/wpp/Excel-Data/population.htm).

Extent of Cyclical Dependence (Research Question 1)

Considering first the total visitor streams (i.e., across all visitor purposes), we find that international tourism has excess volatility compared to the economy in general. The cyclical volatility in tourism numbers relative to the cyclical GDP per capita has an average of 4.96 (see Table 2). Moreover, we find evidence that international tourism behaves in a procyclical manner (Table 3) with a positive (reliability weighted) co-movement elasticity $({\bar{β}}_{i j})$ of 1.27. This implies that the international-tourism sector not only moves in the same direction as the economy as a whole, that is, the co-movement elasticity is larger than zero (meta-analytic Z = 8.12, p < .01) but also that the resulting swings show excess sensitivity, that is, the co-movement elasticity is larger than one (meta-analytic Z = 1.49, p < .10).

Table 2.

Measures of Volatility in the Visitor and Economic Cycles.

	Cyclical Volatility^a,b	Relative Volatility^b,c
Total	0.08	4.96
Holiday	0.11	6.69
Business	0.11	6.04
VFR	0.09	5.34
Economic cycle ( ${GDP}_{j, t}^{c}$ )	0.02	—

Note. GDP = gross domestic product; VFR = visiting friends and relatives.

^aCyclical volatility is the standard deviation of the cyclical component of the visitor or economic series. ^bThe figures for (relative) volatility are weighted averages of the 30 relative volatility measures for the individual countries. The weights are the inverse of the standard errors, normalized to one, where we use bootstrapping to obtain an estimate of these standard errors.

^cRelative volatility is the standard deviation of the cyclical component of the visitor series relative to the standard deviation of the cyclical component of the economic series (GDP/capita).

Table 3.

Synchronization of Visitor Cycle (Co-Movement Elasticity, ${\hat{β}}_{i j}$ ).

		Compared to Zero (= No Co-Movement)		Compared to One (= Excess Co-Movement)
	Effect Size^a	Meta-Analytic Z	Meta-Analytic p value	Meta-Analytic Z	Meta-Analytic p value
Total	1.27	8.12	0.00	1.49	.07
Holiday	1.50	6.86	0.00	2.16	.02
Business	1.06	5.88	0.00	0.29	.39
VFR	0.72	4.44	0.00	−1.71	.96

Note. VFR = visiting friends and relatives.

^aThe effect size is a weighted average of the 30 co-movement elasticities for the individual countries. The weights are the inverse of the standard errors of the co-movement elasticities, normalized to one.

Variability in Cyclical Sensitivity Across Visitor Purposes (Research Question 2a)

The results per visitor purpose show a relative volatility of 6.69 (holiday), 6.04 (business), and 5.34 (VFR), as detailed in Table 2. For the co-movement elasticity, the excess sensitivity observed in the total-visitor series appears to be driven by the holiday visitors, with an average of 1.50, which is significantly higher than one (meta-Z = 2.16, p < .05). We find an almost one-on-one correspondence between the business-cycle fluctuations in the economy as a whole and the number of business visitors, with an average of 1.06, whereas it is 0.72 for VFR. We also test whether the differences between visitor purposes are significant and find that this is not the case for holiday versus business, and business versus VFR, but we do find a significantly smaller effect size for VFR (0.72) than for holiday visitors (1.50): meta-Z = 1.71, p < .10. Because of the personal dimension involved, a VFR trip is less discretionary, making people less likely to postpone or cancel their trip because of unfavorable economic conditions than when booking holiday trips.

Variability in Cyclical Sensitivity Across Countries of Origin (Research Question 2b)

Considerable variability exists in the extent of cyclical dependence across the different countries. While the average cyclical volatility in tourism numbers relative to the cyclical GDP per capita is 4.96, some countries show relative-volatility values below 3 (e.g., Australia and the Netherlands) and others above 10 (e.g., China and Brazil). We also observe a lot of heterogeneity in countries’ extent of synchronization of their tourist numbers to New Zealand with their respective business cycle. While the average co-movement elasticity across countries is 1.27, some countries exhibit a much weaker co-movement elasticity, even below 1 (e.g., again Australia and the Netherlands), while for others it exceeds 1 (e.g., again China and Brazil). We elaborate on the implications of this heterogeneity in the Discussion section.

Asymmetry in the Speed of Adjustment Across Economic Contractions and Expansions (Research Question 3)

We find no evidence for an asymmetric speed of adjustment (see Table 4), neither individually (none of the 120 visitor series showed a significantly negative asymmetry value) nor meta-analytically (Z _total = −0.30, p = .38; Z_holiday = −0.74, p = .23; Z_business = −0.29, p = .39; Z_VFR = 0.35, p = .64). This confirms earlier, untested contentions that the tourism industry is quick to rebound from economic woes (see, e.g., Caterer and Hotelkeeper 2012). For the holiday visitors, this result is consistent with the finding in Oppewal et al. (2010) that many consumers opt to spend a considerable part of sudden increases in their financial means on traveling. As for business visitors, companies may sense that face-to-face contact with their business partners is crucial to quickly pick up speed once the economy starts to show signs of recovery, consistent with recent observations in the business press (U.S. Travel Association 2013).

Table 4.

Asymmetry in Visitor Cycle (Steepness, ST _ij ).

		Compared to Zero (= No Asymmetry)
	Effect Size^a	Meta-Analytic Z	Meta-Analytic p value
Total	−0.04	−0.30	.38
Holiday	−0.10	−0.74	.23
Business	−0.03	−0.29	.39
VFR	0.04	0.35	.64

Note. VFR = visiting friends and relatives.

^aThe effect size is a weighted average of the 30 steepness measures for the individual countries. The weights are the inverse of the standard errors of the steepness measure, normalized to one.

Long-Run Growth Implications (Research Question 4)

Table 5 considers the impact on the sector’s long-run growth prospects of (i) the relative volatility in the number of visitors, (ii) the co-movement elasticity, and (iii) the extent of steepness asymmetry. Across all regressions, the VIF statistics are lower than the common cutoff value of 5, suggesting that multicollinearity is not a concern. While we do not find significant effects for the co-movement and asymmetry statistics, we do obtain a strong significant effect for the relative volatility. This implies that it is not the synchronization per se, nor a differential speed of adjustment, but rather the intrinsic size of the cyclical swings that is linked to growth. When motivating Research Question 4, we have presented arguments for a negative, positive, and a null effect. Empirically, we consistently find a positive and very significant effect for the relative volatility measure (γ_{1, total} = 0.006; γ_{1, holiday} = 0.006; γ_{1, business} = 0.002; γ_{1, VFR} = 0.004; all p < .01).

Table 5.

Models for Long-Run Growth.

	Total Visitors			Holiday Visitors			Business Visitors			VFR Visitors
	Coefficient	p value	VIF	Coefficient	p value	VIF	Coefficient	p value	VIF	Coefficient	p value	VIF
Constant	0.014	0.219	—	−0.008	0.715	—	−0.007	0.398	—	0.023	0.147	—
Relative visitor volatility	0.006	0.001	1.02	0.006	0.022	1.04	0.002	0.001	1.08	0.004	0.006	1.35
Co-movement	0.000	0.970	1.48	0.000	0.915	1.57	0.001	0.634	1.04	0.001	0.567	1.12
Steepness asymmetry in economic cycle	0.000	0.923	1.06	−0.002	0.685	1.03	0.000	0.909	1.12	0.001	0.638	1.01
Growth rate GDP	0.964	0.033	1.62	1.316	0.039	1.71	0.859	0.003	1.20	0.910	0.026	1.40
Growth rate population	0.680	0.335	1.16	0.382	0.756	1.08	1.266	0.004	1.15	0.990	0.268	1.06
R ²	.6185			.5832			.6118			.7388

Note. Bold and underlined indicates significance at 5% (two-sided). GDP = gross domestic product; VFR = visiting friends and relatives; VIF = variance inflation factor.

This finding is in line with the creative destruction argument: More volatility leads to more long-term growth. Volatility seems to serve as a wake-up call for the sector to devote more attention and resources to those countries leading to growth acceleration, while steady settings could lead to complacency.

Comparison to Other Sectors

These empirical findings suggest considerable deviations from earlier findings on nonservice industries. For example, private-label market shares evolve countercyclically (Lamey et al. 2007, 2012), as opposed to the procyclical behavior found here for international tourism. Private-label shares behave asymmetrically across contractions and expansions, as do consumer-durable sales (Deleersnyder et al. 2004), as opposed to the symmetry we document. National brands, in turn, suffer persistent share losses following a downturn, as opposed to the long-term growth increase found here. Similar to the durable-goods sector (Deleersnyder et al. 2004), we find excess (procyclical) sensitivity, albeit with a smaller (average) co-movement elasticity (1.27 vs. 2.25 for consumer durables). These contrasts confirm Srinivasan, Lilien, and Sridhar (2011)’s contention, as well as our conceptual discussion, that service sectors react differently to business-cycle fluctuations than nonservice sectors, and therefore deserve separate research attention.

Robustness Checks

Dynamic Co-Movement

In Equation 1, we do not explicitly allow for dynamic effects, although we allow for serially correlated errors. As a robustness check, we considered the need to augment Equation 1 with a one-period lead $({GDP}_{j, t + 1}^{c})$ and lag $({GDP}_{j, t - 1}^{c})$ term (cf. Deleersnyder et al. 2004). Neither the lead nor the lagged effect is meta-analytically significant for the total-visitor series (meta-analytic Z = 0.23, p = .41 for the lead effect and Z = −1.69, p = .95 for the lagged effect).

Of the six meta-analytic results (three purposes times lead vs. lag), only the lead and lag effect for business visitors is meta-analytically significant (Z = 1.85 and Z = −2.26, respectively, p < .05). Importantly, incorporating the significant lags/leads into a dynamic co-movement elasticity for this visitor purpose does not alter our earlier insight in direction, size, or significance. In particular, the effect size for business visitors becomes 1.16 instead of 1.06.

Accounting for the Economic Situation in the Target Country

The economic situation in the destination country may affect overseas tourism demand through its effect on price levels and exchange rates (cf. Song, Witt, and Li 2009). Table 6 shows the impact of adding the exchange rate and the price index differential between that country and the countries of origin (i.e., $\frac{{CPI}_{NZ, t}}{{CPI}_{j, t}}$ ) as additional control variables in the regressions for long-term growth.⁹ The results of the key variables are very robust, and both additional control variables turned out to be insignificant (p > .10).

Table 6.

Models for Long-Run Growth with Additional Control Variables.

	Total Visitors			Holiday Visitors			Business Visitors			VFR Visitors
	Coefficient	p value	VIF	Coefficient	p value	VIF	Coefficient	p value	VIF	Coefficient	p value	VIF
Constant	0.016	0.178	—	−0.006	0.784	—	−0.006	0.518	—	0.026	0.125	—
Relative visitor volatility	0.006	0.002	1.02	0.006	0.016	1.04	0.002	0.004	1.08	0.004	0.013	1.35
Co-movement	−0.002	0.435	1.48	−0.003	0.503	1.57	0.001	0.731	1.04	0.001	0.737	1.12
Steepness asymmetry in economic cycle	0.000	0.935	1.06	−0.003	0.584	1.03	0.000	0.973	1.12	0.002	0.608	1.01
Growth rate GDP	1.088	0.015	1.62	1.648	0.016	1.71	0.885	0.004	1.20	0.940	0.023	1.40
Growth rate population	0.403	0.605	1.16	−0.072	0.961	1.08	1.308	0.007	1.15	0.854	0.370	1.06
Growth rate CPI_NZ/CPI _j	−0.049	0.894	1.61	−0.207	0.756	1.35	−0.106	0.627	1.32	0.003	0.995	1.39
Growth rate exchange rate	0.022	0.943	1.62	−0.261	0.671	1.39	−0.105	0.578	1.36	0.080	0.829	1.48
R ²	.7167			.5192			.6163			.6761

Note. Bold and underlined values indicate significance at 5% (two-sided). CPI = consumer price index; GDP = gross domestic product; VFR = visiting friends and relatives; VIF = variance inflation factor.

We also consider the addition of the cyclical component in New Zealand’s GDP $({GDP}_{NZ, t}^{c})$ as an additional covariate to Equation 1. For some countries, the cyclical GDP components for the destination and origin country are correlated. For example, the correlation between the cyclical component in the GDP series of Australia and New Zealand is .52. We therefore opted to include (in the spirit of Batra et al. 2000) the part of ${GDP}_{N Z, t}^{c}$ not yet accounted for by the cyclical component in the GDP series of the home country, which we obtained as the residual of an auxiliary regression of ${GDP}_{N Z, t}^{c}$ on ${GDP}_{j, t}^{c}$ . For the total visitors, for example, the economic cycle in the destination country had a significant (p < .05) effect in only six (of 30) instances. Similar to the earlier check for lead and lag effects, we test whether the meta-analytic result is significant, and if that is the case, whether the effect sizes are robust. Even though the meta-analytic results for the co-movement with the domestic business cycle are significant (p < .01), the focal inferences on the impact of the country-of-origin’s business cycle are very robust, with an effect size of 1.27 (total), 1.49 (holiday), 1.03 (business), and 0.73 (VFR) after correcting for the significant ${GDP}_{NZ, t}^{c}$ effects. These are very similar to the originally reported values of 1.27, 1.50, 1.06, and 0.72.

Asymmetric Co-Movement Elasticity

The co-movement elasticity assumes symmetry across the business cycle (Deleersnyder et al. 2004, 2009). We also implement an asymmetric equivalent, allowing for a different co-movement elasticity between expansion and contraction periods. Following Lamey et al. (2012), we define a contraction dummy, which is set to one when the economy in the origin country is contracting (i.e., $Δ {GDP}_{i t}^{c} \leq 0$ ), and zero when it is expanding ( $Δ {GDP}_{i t}^{c} > 0$ ). The interaction of the economic cycle and the contraction dummy is not significant for any of the total visitors’ series, and significant in only 1 of the 90 instances for the individual visitor purposes. Importantly, the meta-analyses on these interaction effects are not significant (Z _total = 0.75; Z _holiday = −0.36; Z _business = 0.67, and Z _VFR = −0.02; p > .10).

Interaction With Time Trend

Some researchers (e.g., De Nardi, French, and Benson 2012) have argued that the service sector has been hit more than usual in the recent GFC. This could be part of a systematic increase over time in the cyclical sensitivity of the service sector. To test for this possibility, we augment Equation 1 with an interaction term between the home country’s cyclical GDP component and a deterministic trend. Meta-analytically, this trend is only significant for business visitors (Z = 3.39, p < .01). Allowing for the significant interactions in the calculation of the effect size for the corresponding co-movement elasticity, we obtain the same effect (1.06) as before when evaluated in the middle of the sample period.

Generalizability to Other Target Countries

To assess whether the substantive insights are idiosyncratic to New Zealand as a destination country, we obtained comparable information for two other target countries: Australia and Japan. Similar to New Zealand, these countries’ island nature facilitates an accurate measurement of the number of international arrivals. However, they differ from New Zealand on a number of important dimensions, such as their size (i.e., Australia with 7.7 M km² is much larger than Japan’s 0.38 M km² and New Zealand’s 0.27 M km²), relative importance in the global economy (Japan is the 3rd economy with a total GDP of 4.9 trillion US$, while Australia and New Zealand are the 12th and 53rd economy, respectively, with a GDP of 1.5 and 0.2 trillion US$; statistics of the UN in 2013), and relative contribution of Travel and Tourism to their economy (cf. Table 1).

For Australia, we purchased visitor data from the Australian Bureau of Statistics. We consider the same time span as for New Zealand, that is, 34 years (1980–2013), and focus on 30 countries of origin, which together accounted for 94% of the international visitors to Australia. The Australian Bureau of Statistics uses the same criteria to classify visitor purposes as Statistics New Zealand. For Japan, we use data from Tourism Japan (http://www.tourism.jp/en/statistics/inbound/). We gathered data of the 30 most visiting countries, which together account for 96% of its international visitors. For Japan, the time span of the data is somewhat shorter, 1990–2013 (24 years). For seven countries, we lack the first year (resulting in a sample period from 1991 to 2013), and for two countries the first 3 years (1993–2013). Next to the total number of visitors, Tourism Japan keeps track of holiday and business visitors. We calculate the number of VFR visitors as the balance of total visitor numbers, after subtracting the number of business and holiday visitors.

Despite the differences between the three countries, we find considerable consistency in the substantive insights, as detailed in Table 7: (i) The average co-movement elasticity for the total visitor series is positive in all three countries, and significantly larger than one in both Japan and New Zealand. For Australia, the weighted mean is also larger than one but not significantly so. (ii) The VFR purpose is the least cyclically sensitivity in all three instances (0.72 for New Zealand, 0.48 for Australia, and 0.40 for Japan), in line with our theorizing. The average co-movement elasticity for holiday visitors is lower in Japan (1.14) than in New Zealand (1.50), while the opposite holds for business visitors (2.57 for Japan vs. 1.06 for New Zealand). This is consistent with Japan being a more business-focused travel destination (with huge potential for ups and downs) and New Zealand a more holiday-focused destination (with huge potential for ups and downs along that dimension). (iii) The relative volatility is considerably larger than one across all three countries and visitor purposes. (iv) In each of the three countries, we find very little evidence of asymmetry, with only one significant effect across the 12 series. Finally (v), the significant positive effect of the relative volatility on the long-run growth rate is replicated for Australia. The effect for Japan, while also positive, does not reach significance.

Table 7.

Comparison Across Destination Countries.

		New Zealand^a		Australia		Japan^c
		Cyclical Volatility	Relative Volatility	Cyclical Volatility	Relative Volatility	Cyclical Volatility	Relative Volatility
Volatility	Total	0.08	4.96	0.08	4.28	0.10	6.38
	Holiday	0.11	6.69	0.10	5.82	0.15	9.32
	Business	0.11	6.04	0.09	5.16	0.07	4.49
	VFR	0.09	5.34	0.08	4.64	0.06	4.20
	Economic cycle	0.02	1.00	0.02	1.00	0.02	1.00
		Effect size^b	Meta-analytic p value (compared to one)	Effect size^b	Meta-analytic p value (compared to one)	Effect size^b	Meta-analytic p value (compared to one)
Synchronization (co-movement)	Total	1.27	.07	1.14	.21	1.52	.01
	Holiday	1.50	.02	1.42	.02	1.14	.30
	Business	1.06	.39	1.60	.00	2.57	.00
	VFR	0.72	.96	0.48	1.00	0.40	1.00
		Effect size^b	Meta-analytic p value (compared to zero)	Effect size^b	Meta-analytic p value (compared to zero)	Effect size^b	Meta-analytic p value (compared to zero)
Asymmetry (steepness)	Total	−0.04	.38	− 0.21	.07	−0.13	.23
	Holiday	−0.10	.23	−0.11	.20	−0.17	.19
	Business	−0.03	.39	−0.12	.13	−0.08	.28
	VFR	0.04	.64	−0.05	.34	−0.21	.11
		Coefficient	p value	Coefficient	p value	Coefficient	p value
Long-term effect relative volatility	Total	0.006	.001	0.006	.047	0.001	.425
	Holiday	0.006	.022	0.005	.125	0.001	.685
	Business	0.002	.001	0.003	.010	0.002	.335
	VFR	0.004	.006	0.007	.004	0.000	.978

Note. Bold and underlined values indicate significance at 5%; Italic and underlined indicates significance at 10% (one-sided for the meta-analytic results of co-movement and steepness and two-sided for the regression parameters of relative volatility). VFR = visiting friends and relatives.

^aResults for New Zealand are also reported in previous tables: volatility (Table 2), co-movement (Table 3), steepness (Table 4), and long-term effect of relative volatility (Table 5). ^bThe effect sizes are weighted averages of the 30 measures for the individual countries. The weights are the inverse of the standard errors, normalized to one. ^cVFR for Japan is the balance of total visitor numbers, after subtracting the number of business and holiday visitors.

Accounting for the Cyclical Dependence in Marketing Spending

Governments regularly invest in tourism marketing to promote their country as an attractive tourism destination. Big spenders in this respect include the United Kingdom (US$160 million) and the United States (US$150 million; The New York Times 2012). Also New Zealand has a rich history with tourism marketing and spent NZ$65 million in 2012 (Tourism New Zealand 2013), which is approximately US$53 million.

In a final robustness check, we assess to what extent cyclical variations in New Zealand’s total tourism marketing budget¹⁰ affect the tourist numbers’ co-movement elasticity. To that extent, we added the cyclical component of the total tourism promotion series to Equation 1. We find that this has a significant meta-analytic effect in all four instances (Z _total = 8.92, Z _holiday = 8.69, Z _business = 2.85, and Z _VFR = 4.55; p < .01). However, accounting for the significant marketing effects does not alter the inferences on the focal co-movement elasticities, with effect sizes (1.43, 1.59, 1.07, and 0.80 for the total series, holiday visitors, business visitors, and VFR) that remain similar to the ones reported before (1.27, 1.50, 1.06, and 0.72).

Discussion

In spite of the growing importance of the service industry, and in spite of its very distinct features that prevent a direct transfer of findings from these other sectors, little is known on how service demand behaves across economic expansions or contractions, or on how this cyclical dependence affects its future growth prospects. Focusing on one of the most important export services, tourism, we systematically investigate how business-cycle fluctuations affect international visitor streams. Moreover, unlike previous studies, which considered domestic demand only and the corresponding business cycle (Deleersnyder et al. 2004; Gordon, Goldfarb, and Li’s 2013; van Heerde et al. 2013), the very nature of the service studied (international tourism) necessitates the consideration of multiple business cycles, which in themselves differ widely in intensity and variability.

We analyze several decades of data, covering multiple expansion and contraction periods, across different countries of origin in various continents, and across different purposes for visiting a destination. The insights from the 100+ international-visitor series are subsequently combined meta-analytically in a novel set of empirical regularities on the link between two important phenomena: cyclical fluctuations in the state of the economy and international tourism.

We offer empirical evidence for an excess sensitivity of international tourism to cyclical up- and downswings in the economy of the country of origin. The relative volatility, for the total visitor numbers and for each of the three considered visitor purposes, is consistently larger than one. While the overall co-movement elasticity is larger than one, it varies systematically across visitor purposes. Importantly, it is significantly higher than one for the number of holiday visitors. Given that many economies rely heavily on income from touristic services, such excess sensitivity is especially worrisome in recessionary times when governmental income streams are already under pressure. The co-movement elasticity is close to one for business visitors, but is less than one for VFR, indicating that this type of tourism is least sensitive to the cycle. For countries where the growth in incoming VFR exceeds the growth in holiday visitors, as is the case in the United Kingdom (U.K. Civil Aviation Authority 2009) and New Zealand, this contributes to an increased stability in their future international passenger traffic.

Other developments, however, are expected to result in an increased volatility. Focusing once more on New Zealand, we classify in Table 8 the countries of origin along two dimensions: (i) the size of their co-movement elasticity (smaller or larger than one) and (ii) the extent of the cyclical volatility in the country’s economy (above or below the median value). We do so for the largest segment, holiday visitors.

Table 8.

Cross-Country Heterogeneity in Cyclical Dependence, for Holiday Visitor Numbers.

			Economic Volatility $(σ_{j} ({GDP}_{j, t}^{c}))$
			Low (Below Median Split)	High (Above Median Split)
	Low (Elasticity < 1)	Countries	Cell (1, 1): Australia, Germany, the Netherlands, Switzerland, France, Spain	Cell (1, 2): United Kingdom, Singapore, Hong Kong, Malaysia, Fiji, Ireland
		Average co-movement	−0.451	0.346
		Average economic volatility	0.013	0.023
		(Weighted) average LT growth^a	0.049	0.070
Co-movement Elasticity ${(β}_{total, j})$		Market share (1980–2011)	38.5%	15.6%
	High (Elasticity > 1)	Countries	Cell (2, 1): United States, Japan, Canada, South Africa, India, Sweden, Denmark, Italy, Austria	Cell (2, 2): South Korea, China, Taiwan, Thailand, Samoa, Tonga, Indonesia, Brazil, Philippines
		Average co-movement	1.803	4.638
		Average economic volatility	0.014	0.025
		(Weighted) average LT growth^a	0.032	0.140
		Market share (1980–2011)	32.2%	13.8%

Note. GDP = gross domestic product; LT = long-term.

^aWeighed based on the within-cell share of each country.

The New Zealand tourist authorities will obviously appreciate that the low-risk cell (1, 1) commands the largest market share (38.5%), compared to 13.8% for the highest risk (2, 2) cell. Visitor streams from countries in the highest risk (2, 2) cell are vulnerable along both dimensions: Their economy in itself shows pronounced cyclical swings, with a $σ_{j} ({GDP}_{j, t}^{c})$ almost twice as large (0.025 on average) than the countries in the low-risk (1, 1) cell (0.013), while their tourist numbers are also very sensitive to these swings, with an average co-movement elasticity of 4.64. However, the (market share) weighted long-run growth (i.e., the slope in ${Visitors}_{i j, t}^{LR}$ ) is more than twice as large in the highest risk (2, 2) cell. Because of this, the country can expect that the changing composition of its visitor base (with increasing numbers coming from countries such as China, Brazil, and Indonesia) will entail a higher cyclical volatility as well. These findings underscore the need to closely monitor not only the current composition but also the expected evolution in the various components of a country’s visitor base.

Given that there is considerable variability across the different countries of origin, various diversification opportunities exist. To maintain their share in an increasingly competitive market, private and public organizations spend substantial amounts of money to promote their country as an attractive destination (Kulendran and Dwyer 2009). A judicious allocation of these funds, longitudinally as well as across countries, can help not only to increase the overall number of visitors but also to attenuate the resulting volatility. While the visitor-number objective is reflected in allocation-support systems (see, e.g., Mazanec 1986 or Wierenga 1981), the potential to reduce the overall volatility has, thus far, been ignored in the literature. Better insights in the composition of, and evolution in, a country’s visitor base are also crucial to better tailor the accommodations and services. For example, visitors from certain countries (e.g., Japan) are known to spend shorter periods of time in a country compared to North American and European visitors and to focus more on a limited set of major attractions within a country (Page 1989). In times where the number (proportion) of Japanese visitors is expected to peak, sufficient accommodations with Japanese-speaking personnel near the major destinations are called for, and restaurants may consider to adjust their menus to better accommodate these visitors’ tastes. Similarly, service providers may want to emphasize the short transit times between locations to attenuate the negative evolution in visitor numbers that can be expected when the Japanese economy slows down.

While both tourist organizations and governments will deplore the high sensitivity of international tourism to economic downturns, the results also offer two pieces of favorable news. First, unlike the economy as a whole and numerous individual sectors that recover slowly following a more rapid decline, we find no evidence of such asymmetric adjustment patterns in this service sector. Hence, international tourism is quicker to rebound following an economic downturn than many other industries and could therefore serve as an early indicator of (and engine for) a subsequent recovery in other sectors.

Importantly, we find no evidence that the excess sensitivity hurts the long-run growth prospects of the sector. For two of the considered destination countries, we find that a higher relative volatility has a positive impact on the long-run growth rate in the number of visitors coming from a given country, and this for each of the different visitor purposes.

Limitations and Future Research

This research has some limitations that offer avenues for further research. While we consider multiple countries of origin (30) and various visitor purposes, resulting in a rich set of empirical regularities, we focus on only three destination countries. It would be interesting to replicate this study with other target countries, also from other continents. In addition, we focus on the number of international visitors. However, economic up- and downturns may also affect the length of stay and the amount of expenditures while traveling. It would be of interest to also study the cyclical sensitivity of these other metrics.

More broadly, even though we focus on a very important service sector, we should keep in mind that the service industry in itself is not a single, homogeneous group of activities. Other services may react in a different way to economic contractions and expansions. For example, domestic destinations may well become more popular in economically difficult times to substitute for the more expensive international trips. Also, some services (such as haircuts, doctors’ services, or tax-accounting services) are intrinsically less discretionary and more difficult to postpone, than big-ticket items such as international tourism, and may be driven more by the domestic, rather than the foreign, business cycle.

Therefore, a key area in need of more research is to better explore the heterogeneity in cyclical sensitivity within the service industry. For example, different services can be more or less discretionary, more or less capital-intensive to produce and/or consume, characterized by a different income elasticity, or be more or less sensitive to social-visibility considerations. These factors may affect the severity of the cutbacks during recessionary times, the speed of recovery during expansions, or the long-run growth rate following cyclical swings. A better understanding of the relative importance of these contextual factors would not only add to the generalizability of the insights but also aid further theory development on the importance of macroeconomic developments to service firms.

Relatedly, it would be useful to study the role of managerial activities in moderating the impact of the economic tides on various services. Do customers consistently value certain service attributes differently in economic upturns versus downturns, as recently proposed in Kumar et al. (2014)? Should firms adjust their pricing scheme and spend more or less on advertising during a recession? Given that many competitors cut their advertising support, there will be less clutter in the marketplace, which could increase the effectiveness. Alternatively, customers may be less receptive to advertising during economically difficult times. Also, should firms adjust their portfolio of services offered in function of the economic climate, and how should firms prepare their frontline employees to deal with customers who become more price sensitive when the going gets tough, or who are thinking of postponing certain repeat visits longer? Clearly, the moderating role of marketing conduct at various levels of the service organization offers exciting opportunities for new research.

Footnotes

Authors’ Note

This article was partly written while the first author was visiting Singapore Management University as Tommy Goh Visiting Professor in Entrepreneurship and Business.

Acknowledgments

The authors gratefully acknowledge suggestions from the participants at the 2013 Marketing Dynamics Conference.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors gratefully acknowledge financial support from the New Zealand Royal Society Marsden Fund (MAU1012).

Notes

References

Ailawadi

Kusum

Pauwels

Koen

Steenkamp

J. B.

(2008), “Private Label Use and Store Loyalty,” Journal of Marketing, 72 (November), 19–30.

Allenby

Greg M.

Jen

Lichung

Leone

Robert P.

(1996), “Economic Trends and Being Trendy: The Influence of Consumer Confidence on Retail Fashion Sales,” Journal of Business and Economic Statistics, 14 (January), 103–111.

Backer

Elisa

(2012), “VFR Travel: It Is Underestimated,” Tourism Management, 33 (February), 74–79.

Balaguer

Jacint

Cantavella-Jordá

Manuel

(2012), “Tourism as a Long-Run Economic Growth Factor: The Spanish Case,” Applied Economics, 34 (7), 877–884.

Batra

Rajeev

Ramaswamy

Venkatram

Alden

Dana L.

Steenkamp

Jan-Benedict E. M.

Ramachander

(2000), “Effects of Brand Local and Nonlocal Origin on Consumer Attitudes in Developing Countries,” Journal of Consumer Psychology, 9 (2), 83–95.

BusinessWeek (2009), “Global Travel Takes a Dive,” August 4 (accessed February 2, 2013), [available at http://www.businessweek.com./globalbiz/content/aug2009/gb2009084_138150.htm].

Caballero

Ricardo J.

(1993), “Durable Goods: An Explanation for Their Slow Adjustment,” Journal of Political Economy, 101 (April), 351–384.

Caterer and Hotelkeeper (2012), “International Tourism Proves Resilient Despite Tough Economic Environment,” November 6 (accessed February 2, 2013), [available at http://www.catererandhotelkeeper.co.uk/articles/6/11/2012/346119/international-tourism-proves-resilient-through-economic-environment.htm].

Cerqueira

Pedro A.

(2013), “A Closer Look at the World Business Cycle Synchronization,” International Economics and Economic Policy, 10 (September), 349–363.

10.

Christiano

Lawrence J.

Fitzgerald

Terry J.

(1998), “The Business Cycle: It’s Still a Puzzle,” Federal Reserve Bank of Chicago Economic Perspectives, 22 (December), 56–83.

11.

Cook

Steven

(1999), “Cyclicality and Durability: Evidence from U.S. Consumers’ Expenditure,” Journal of Applied Economics, 11 (November), 299–310.

12.

Coviello

Nicole

Winklhofer

Heidi

Hamilton

Karla

(2006), “Marketing Practices and Performance of Small Service Firms—An Examination in the Tourism Accommodation Sector,” Journal of Service Research, 9 (August), 38–58.

13.

Cuadrado-Roura

Juan R.

O. V.-Abarca

Alvaro

(2001), “Business Cycle and Service Industries: General Trends and the Spanish Case,” The Services Industries Journal, 21 (January), 103–122.

14.

Deleersnyder

Barbara

Dekimpe

Marnik G.

Steenkamp

Jan-Benedict E. M.

Leeflang

Peter H.

(2009), “The Role of National Culture in Advertising’s Sensitivity to Business Cycles: An Investigation across Continents,” Journal of Marketing Research, 46 (October), 623–636.

15.

Deleersnyder

Barbara

Dekimpe

Marnik G.

Sarvary

Miklos

Parker

Philip M.

(2004), “Weathering Tight Economic Times: The Sales Evolution of Consumer Durables over the Business Cycle,” Quantitative Marketing and Economics, 2 (December), 347–383.

16.

De Nardi

Mariacristina

French

Eric

Benson

David

(2012), “Consumption and the Great Recession,” Federal Reserve Bank of Chicago Economic Perspectives, 36 (Spring), 1–16.

17.

Dobbs

Richard F.

Karakolev

Tomas

Maligne

Francis

(2002), “Learning to Love Recessions,” McKinsey Quarterly, 2 (Special Edition), 6–9.

18.

Dolnicar

Sara

Crouch

Geoffrey I.

Devinney

Timothy

Huybers

Twan

Louviere

Jordan J.

Oppewal

Harmen

(2008), “Tourism and Discretionary Income Allocation—Heterogeneity Among Households,” Tourism Management, 29 (February), 44–52.

19.

Döpke

Jörg

(2004), “How Robust Is the Empirical Link Between Business-Cycle Volatility and Long-Run Growth in OECD Countries?” International Review of Applied Economics, 18 (January), 103–121.

20.

European Network for Accessible Tourism (2009), “Economic Recession—Time for Renewal in the European Tourism Sector,” EFFAT, (accessed February 21, 2013), [available at http://www.accessibletourism.org].

21.

Engle

Robert F.

(1974), “Band Spectrum Regression,” International Economic Review, 15 (February), 1–11.

22.

Farebrother

R. W.

(1980), “Algorithm AS 153: Pan’s Procedure for the Tail Probabilities of the Durbin-Watson Statistic,” Journal of the Royal Statistical Society. Series C (Applied Statistics), 29 (2), 224–227.

23.

Falvey

Rodney E.

Gemmell

Norman

(1996), “Are Services Income-Elastic? Some New Evidence,” Review of Income and Wealth, 42 (September), 257–269.

24.

Filardo

Andrew J.

(1995), “Recent Evidence on the Muted Business Cycle,” Federal Reserve Bank of Kansas City Economic Review, 80 (Q II), 27–43.

25.

Filardo

Andrew J.

(1997), “Cyclical Implications of the Declining Manufacturing Employment Share,” Federal Reserve Bank of Kansas City Economic Review, 82 (Q II), 63–87.

26.

Frechtling

Douglas C.

(1982), “Tourism Trends and the Business Cycle: Tourism in Recession,” Tourism Management, 1 (December), 285–290.

27.

Gordon

Brett R.

Goldfarb

Avi

Yang

(2013), “Does Price Elasticity Vary with Economic Growth? A Cross-Category Analysis,” Journal of Marketing Research, 50 (February), 4–23.

28.

Guyt

Jonne Y.

Gijsbrechts

Els

(2014), “Take Turns or March in Sync? The Impact of the National Brand Promotion Calendar on Manufacturer and Retailer Performance,” Journal of Marketing Research, 51 (December), 753–772.

29.

Hodrick

Robert J.

Prescott

Edward C.

(1997), “Postwar U.S. Business Cycles: An Empirical Investigation,” Journal of Money, Credit and Banking, 29 (February), 1–16.

30.

Holmqvist

Jonas

Grönroos

Christian

(2012), “How Does Language Matter for Services? Challenges and Propositions for Service Research,” Journal of Service Research, 15 (November), 430–442.

31.

Kamakura

Wagner A.

Y. Du

Rex

(2012), “How Economic Contractions and Expansions Affect Expenditure Patterns,” Journal of Consumer Research, 39 (August), 229–247.

32.

Katona

George

(1975), Psychological Economics. New York: Elsevier Scientific.

33.

Kramer

Roderick M.

(2002), “When Paranoia Makes Sense,” Harvard Business Review, 80 (July), 62–69.

34.

Kulendran

Nada

Dwyer

Larry

(2009), “Measuring the Return from Australian Tourism Marketing Expenditure,” Journal of Travel Research, 47 (February), 275–284.

35.

Kumar

Umashankar

Nita

Kim

Kihyun H.

Bhagwat

Yashoda

(2014), “Assessing the Influence of Economic and Customer Experience Factors on Service Purchase Behaviors,” Marketing Science, 33 (September-October), 673–692.

36.

Lamey

Lien

Deleersnyder

Barbara

Dekimpe

Marnik G.

Steenkamp

Jan-Benedict E. M.

(2007), “How Business Cycles Contribute to Private-Label Success: Evidence from the U.S. and Europe,” Journal of Marketing, 71 (January), 1–15.

37.

Lamey

Lien

Deleersnyder

Barbara

Steenkamp

Jan-Benedict E. M.

Dekimpe

Marnik G.

(2012), “The Effect of Business Cycle Fluctuations on Private-Label Share: What Has Marketing Conduct Got to Do with It?,” Journal of Marketing, 76 (January), 1–19.

38.

Lebow

David E.

Sichel

Daniel E.

(1995), “Is the Shift Toward Employment in Services Stabilizing,” in The Service Productivity and Quality Challenge, Harker

Patrick T.

, ed. Dordrecht, the Netherlands: Kluwer Academic Publishers, 81–112.

39.

Lovelock

Christopher H.

(1997), “Fear of a Recession,” Marketing Management, 6 (Fall), 14–16.

40.

Mazanec

Josef A.

(1986), “A Decision Support System for Optimizing Advertising Policy of a National Tourist Office—Model Outline and Case Study,” International Journal of Research in Marketing, 3 (June), 63–77.

41.

Mehra

Yash P.

(2001), “The Wealth Effect in Empirical Life-Cycle Aggregate Consumption Equations,” Federal Reserve Bank of Richmond Economic Quarterly, 87 (Spring), 45–68.

42.

Moran

Robert T.

Harris

Philip R.

Moran

Sarah V.

(2007), Managing Cultural Differences: Global Leadership Strategies for the 21st Century, 7th ed. Oxford, UK: Elsevier.

43.

Nijs

Vincent R.

Srinivasan

Shuba

Pauwels

Koen

(2007), “Retail-Price Drivers and Retailer Profits,” Marketing Science, 26 (July-August), 473–487.

44.

Nooteboom

Bart

Berger

Hans

Noorderhaven

Niels G.

(1997), “Effect of Trust and Governance on Relational Risk,” Academy of Management Journal, 40 (2), 308–338.

45.

Oppewal

Harmen

Paas

Leonard J.

Crouch

Geoffrey I.

Huybers

Twan

(2010), “Segmenting Consumers Based on How They Spend a Tax Rebate: An Analysis of the Australian Stimulus Payment,” Journal of Economic Psychology, 31 (August), 510–519.

46.

Y.-C.

de Vries

Lisette

Wiesel

Thorsten

Verhoef

Peter C.

(2014), “The Role of Consumer Confidence in Creating Customer Loyalty,” Journal of Service Research, 17 (August), 339–354.

47.

Page

Stephen

(1989), “Changing Patterns of International Tourism in New Zealand,” Tourism Management, 10 (December), 337–341.

48.

Petev

Ivaylo

Pistaferri

Luigi

Eksten

Itay Saporta

(2012), “An Analysis of Trends, Perceptions, and Distributional Effects in Consumption,” in The Great Recession, Grusky

David B.

Western

Bruce

Wimer

Christopher

, eds. New York: Russell Sage, 161–195.

49.

Rosenthal

Robert

(1991), Meta-Analytic Procedures for Social Research. Newbury Park, CA: Sage.

50.

Rust

Roland T.

Chung

Tuck S.

(2006), “Marketing Models of Service and Relationships,” Marketing Science, 25 (November-December), 560–580.

51.

Schumpeter

Joseph A.

(1939), Business Cycles: A Theoretical, Historical, and Statistical Analysis of the Capitalist Process, 1st ed. New York: McGraw-Hill.

52.

Sichel

Daniel E.

(1993), “Business Cycle Asymmetry: A Deeper Look,” Economic Inquiry, 32 (April), 224–236.

53.

Song

Haiyan

Gang

Witt

Stephen F.

Fei

Baogang

(2010), “Tourism Demand Modelling and Forecasting: How Should Demand be Measured?” Tourism Economics, 16 (March), 63–81.

54.

Song

Haiyan

Witt

Stephen F.

Gang

(2009), The Advanced Econometrics of Tourism Demand. New York: Routledge.

55.

Srinivasan

Raji

Lilien

Gary L.

Sridhar

Shrihari

(2011), “Should Firms Spend More on Research and Development and Advertising During Recessions,” Journal of Marketing, 75 (May), 49–65.

56.

Statistics New Zealand (2012), Tourism Satellite Account: 2012. Wellington: Statistics New Zealand, (accessed Febraury 21, 2013), [available at www.stats.govt.nz].

57.

Steenkamp

Jan-Benedict E. M.

Fang

Eric (Er)

(2011), “The Impact on Economic Contractions on the Effectiveness of R&D and Advertising: Evidence from U.S. Companies Spanning Three Decades,” Marketing Science, 30 (July-August), 628–645.

58.

Stock

James H.

Watson

Mark W.

(1999), “Business Cycle Fluctuations in U.S. Macroeconomic Time Series,” in Handbook of Macroeconomics, Taylor

John

Woodford

Michael

, eds. Amsterdam: Elsevier Science, 3–64.

59.

Tellis

Gerard J.

Tellis

Kethan

(2009), “Research on Advertising in a Recession,” Journal of Advertising Research, 49 (September), 304–327.

60.

The New York Times (2012), “Selling America abroad,” April 2 (accessed February 21, 2013), [available at http://www.nytimes.com/2012/04/03/business/with-brand-usa-us-in-big-drive-to-attract-foreign-tourists.html?_r=0].

61.

The Telegraph (2008), “Financial Crisis Forces Cuts in Business Travel,” October 21 (accessed February 2, 2013), [available at http://www.telegraph.co.uk/travel/travelnews/3235079/Financial-crisis-forces-cuts-in-business-travel.html].

62.

The Wall Street Journal (2009), “The Recession Doesn’t Spare Travel & Tourism,” June 16 (accessed February 2, 2013), [available at http://blogs.wsj.com/economics/2009/06/16/the-recession-doesnt-spare-travel-tourism/].

63.

Tourism New Zealand (2013), Tourism New Zealand Annual Report 2012-2013, (accessed February 21, 2014), [available at http://www.tourismnewzealand.com/about-us/publications/].

64.

U.K. Civil Aviation Authority (2009), “International Relations: The Growth in Air Travel to Visit Friends and Relatives,” (accessed February 21, 2013), [available at http://www.caa.co.uk/publications].

65.

United Nations (2008), International Recommendations for Tourism Statistics. New York: United Nations Publications.

66.

United Nations World Tourism Organization (2012), Tourism Highlights, 2012 ed., (accessed February 21, 2013), [available at http://mkt.unwto.org].

67.

Urquhart

Michael

(1981), “The Services Industry: Is It Recession-Proof?” Monthly Labor Review, 104 (October), 12–18.

68.

U.S. Travel Association (2013), “The Role of Business Travel in the U.S. Economic Recovery,” 32pp, (accessed Frebruary 21, 2014), [available at http://www.ustravel.org].

69.

van Heerde

Harald J.

Gijsenberg

Maarten J.

Dekimpe

Marnik G.

Steenkamp

Jan-Benedict E. M.

(2013), “Price and Advertising Effectiveness over the Business Cycle,” Journal of Marketing Research, 50 (April), 177–193.

70.

Webber

Dominic

Buccellato

Tullio

White

Sean

(2010), “The Global Recession and its Impact on Tourists’ Spending in the UK,” Economic & Labour Market Review,” 4 (August), 65–73.

71.

Wierenga

Berend

(1981), “Modelling Impact of Advertising and Optimizing Advertising Policy: An Application in Recreation,” European Journal of Operational Research, 8 (November), 235–248.

72.

Zeithaml

Valarie

Parasuraman

Ananthanarayanan

Berry

Leonard L.

(1985), “Problems and Strategies in Services Marketing,” Journal of Marketing, 49 (Spring), 33–46.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB