The Impact of Tourism on the Resilience of the Turkish Economy: An Asymmetric Approach

Abstract

The impact of tourism on economic growth is a subject of interest to researchers as well as policy makers. Numerous studies have explored this relationship, often arriving at varying conclusions depending on the methods employed. Most of these studies, however, assume a symmetric relationship between tourism and economic growth. In this study, the Hatemi-J asymmetric causality test was used to test for short-run asymmetric causality between tourism receipts and economic growth in Turkey for the period 1990–2023, and unidirectional causality was found between the increase in tourism incomes and economic growth and between the decrease in tourism incomes and economic contraction. Additionally, the hidden co-integration test was applied to examine the asymmetric relationships between them in the long run, and the results reveal that an increase in tourism revenues provides resilience to the economy by mitigating contraction during economic downturns. This study contributes to the field by addressing the interaction of tourism and the economy in Turkey from an asymmetric perspective and by revealing previously unobserved relationships. The results provide partial support for the tourism-led growth hypothesis. In the long term, it is recommended that policymakers design tourism strategies aimed at enhancing resilience to economic shocks, thereby also strengthening the national economy. Diversified markets and products, well-structured incentives, and sustainable tourism practices are key elements in achieving this goal.

1. Introduction

Tourism is one of the most important components of the global economy. According to UNWTO, international tourist arrivals have roughly doubled from 676 million in 2000 to 1.3 billion in the intervening 23 years, despite many shocking impacts such as the global economic crisis, SARS and COVID-19 outbreaks [1]. Tourism is estimated to generate 3% of global GDP, worth USD 3.4 trillion in 2023 [2]. In addition to this economic size, tourism should also be assessed in terms of its multiplier effect [3,4,5,6]. Many sectors such as transport, agriculture, health and industry are intertwined with tourism. According to World Travel and Tourism Council calculations, the travel and tourism industry created 334 million jobs in 2019 before the pandemic, corresponding to 10.5% of jobs worldwide with direct, indirect, and induced impacts [7].

The strong economic impact of the tourism sector has led many researchers to question the relationship between tourism and economic growth. Today, the sustainability perspective has come to the fore in order for economic growth to produce results that extend to economic development.

In the literature, the concept of sustainable growth is analyzed in three dimensions: environmental, social and economic. The environmental dimension is concerned with achieving growth without compromising the country’s natural resources, while the social dimension emphasizes social equity and quality of life. The economic dimension pertains to the inclusive and stable growth of the system on a sound macroeconomic foundation in the long term [8,9,10]. In this context, one of the important indicators of the success of economic sustainability is the resilience of the system to shocks. In a general definition, resilience refers to the ability of a system to respond to shocks, adapt and recover to reach equilibrium in the long run [11,12]. Although the concepts of sustainability and resilience focus on different aspects, they both consider the resilience and stability of the system in the long run.

One of the success indicators of economic resilience is the stable realization of growth. This study aims to assess the contribution of the tourism sector in Turkey to the goal of sustainable economic growth from the perspective of resilience.

In the literature, the impact of tourism on economic growth is mainly shaped by the tourism-led growth hypothesis. The tourism-led growth hypothesis (TLGH), as an extension of the export-led growth hypothesis (ELGH), has a similar approach that questions the impact of tourism on growth in the short and long run [13,14].

The literature has largely analyzed the relationship between economic growth and tourism from a symmetric perspective. The number of studies analyzing the relationship from an asymmetric perspective using relatively new econometric methods is rather limited. However, while the symmetric view assumes that positive and negative shocks have similar effects, the asymmetric view can treat the effects of positive and negative shocks separately and reveal the hidden relationship that cannot be detected with the symmetric view [15]. In order to learn more about the resilience effect that tourism can have on the economy in Turkey, the relationships are examined with an asymmetric perspective. Although there are various publications that question the relationship between tourism and economic growth asymmetrically, to the best of our knowledge, no study was identified that addresses this issue in the context of resilience, in other words, a study that questions the impact of the tourism sector on economic resilience.

The predominant part of tourism is not a compulsory need for people, it is generally an activity to see new places, to enjoy, to rest, to establish new business contacts. Therefore, this sector has the potential to be rapidly affected by economic factors such as price, exchange rate, transport costs, as well as non-economic determinants such as politics, security, environment, etc., [16,17,18,19]. A political crisis or security concerns between two countries can lead tourists to a third destination in a short period of time [20,21]. Or a health scare, such as the COVID-19 pandemic, can bring the sector to a standstill [22]. This vulnerability to many factors is reflected in the volatility of the sector’s income. In other words, the potential for tourism to face negative or positive shocks is high and the issue needs to be approached from the perspective of asymmetric interactions.

Turkey, on the other hand, has experienced its own unique shocks in terms of the overall economy and the structure of the country’s tourism, and has the potential to do so in the future, making it a good laboratory for observing asymmetric relationships. That is to say, the main tourist destinations are the provinces of Antalya and Mugla. Seventy per cent of overnight stays by foreign tourists take place in these two provinces [23]. Tourism in both destinations is largely dependent on two markets: Russia and Germany. A political or economic problem with one of these markets causes a large negative shock that spreads to tourism in general, followed by a large positive shock once the problem is resolved. In addition, the turmoil and upheaval in the Middle East, Turkey’s immediate neighbor, has caused shocks to the tourism and whole economy in recent years and has the potential to do so again. The events and crises that are considered to have an impact on tourism in Turkey are listed in

Table 1. Events and crises affecting tourism in Turkey.

Year	Crisis or Events
1994	Turkey’s economic crisis Political instability High inflation Currency crisis High borrowing costs of government
1999	Marmara earthquake Political and economic stability Security concerns—terrorist activities 1998 Asian financial crisis
2001	Turkey’s economic crisis Global economic slowdown (dot-com bubble burst) September 11 attacks—security concerns globally Regional instability Middle-East
2006	Health concerns avian flu outbreak Regional instability Middle-East Security concerns—terrorist activities Increased competition (such as Greece, Spain, Croatia) Exchange rate fluctuations
2009	Impact of 2008 global financial crisis Export crisis (esp. to European Union) Reduction in investments/decline in capital flows Decline in domestic demand/industrial production
2010	Global economic uncertainty Low-budget tourism Increased competition (such as Greece, Spain, Egypt) Currency fluctuations (expensive destination)
2016	Security concerns—terrorist activities—failed coup attempt Turkey–Russia tensions Regional instability—Middle East Economic and political uncertainty
2018–2019	Political tension with US High external debt Inflation and current account deficit
2020	COVID-19 pandemic restrictions Economic recession and financial uncertainty Shift in travel behavior to local tourism

Source: Compiled by the authors.

.

The impact of these crisis factors on tourism income and GDP per capita is shown in Figure 1 and Figure 2. Figure 1 shows the evolution of tourism income and GDP values over the years. Figure 2 shows the percentage change in the same variables. The impact of the events listed in Table 1 is clearly visible in both figures.

The structure of the Turkish economy and the importance given to tourism have some unique characteristics. Foreign trade deficit is a chronic problem for the Turkish economy. As a matter of fact, as of 2023, the foreign trade deficit has reached USD 106 billion in an economy with a GDP of USD 1024 billion [24]. In the same period, tourism revenues totaled USD 54.3 billion [25]. In 2013, Morgan Stanley ranked Turkey among the so-called fragile five (Brazil, India, Indonesia, South Africa and Turkey) according to factors including the current account deficit [26,27]. According to the results of Hanke’s Annual Misery Index (HAMI) 2023, which calculates the general course of the country’s economy using variables such as unemployment, inflation, bank lending rates and real GDP per capita, and where high rankings are characterized as negative, Turkey ranked 7th [28]. As a result, the Turkish economy has a significant need for tourism in terms of its contribution to the economy, especially its ability to generate foreign exchange inflows.

In conclusion, both Turkish tourism and the economy as a whole are highly exposed to both positive and negative shocks. This structure provides us with a good laboratory to observe the relationship between tourism and the wider economy from a resilience perspective in times of shocks.

We computed the cumulative negative and positive shocks of tourism incomes and GDP per capita in Turkey. By analyzing long-run relationships through these cumulative shocks and employing the hidden co-integration techniques by Granger and Yoon [15], we tested the asymmetric tourism-led growth hypotheses for Turkey. The hidden co-integration methodology has the advantage of revealing the relationships between the positive and negative cumulative shocks of the series. Moreover, the Granger and Yoon [15] latent cointegration test, which helps to establish short-run relationships between cumulative positive and negative shocks of variables, facilitates the estimation of the subsequent crouching error correction model (CECM). Here, the determinants of long-run dynamics are identified based on the estimation of the CECM.

In summary, the primary contribution of this study lies in the asymmetric testing of the tourism-led growth hypotheses and questioning the results from a resilience perspective. The secondary contribution involves investigating short-run asymmetric causality. This study adds to the ongoing discourse surrounding the relationship between tourism and growth, offering a fresh perspective.

The following section of the research provides a review of the literature concerning the tourism-led growth hypothesis. Section 3 outlines the materials, econometric methodology, and the asymmetric hypotheses. Section 4 presents empirical results and Section 5 presents the discussion. The final section contains concluding remarks.

2. Literature Review

Sustainability is defined as the ability to use available resources in a way that meets the needs of future generations and maintains ecological, social and economic balance [29,30]. While environmental sustainability is concerned with the protection of natural resources and the transfer of the ecosystem to future generations, economic sustainability is concerned with concepts such as social equality, participation and justice [31,32,33].

Resilience, on the other hand, focuses on the ability of systems to survive, adapt and recover from shocks and changes [34,35]. While the theoretical literature on resilience is predominantly ecological, White et al. [36] present definitions of resilience by discipline and state that the concept of economic resilience is dealt with under the main headings of dynamic, static, inherent and adaptive. In this sense, the resilience of the economy is considered as an important factor in achieving stable and sustainable development [37].

The relationship between these two concepts has been widely questioned in the literature in a number of areas. Despite some conflicts in specific situations, as noted by Carpenter et al. [38], there is a large body of literature suggesting that the two concepts are complementary and that there is a strong relationship between them [8,10,31,35,39,40]. The resilience of the economy is considered to be an important factor in achieving stable and sustainable development [9]. Weber [10] considered resilience to be a component of sustainability and emphasized that the two concepts are interdependent.

An analysis of measurement indicators can help to understand the differences and similarities between these concepts. Sustainability measurement is typically based on indicators such as carbon footprint, biodiversity loss, energy intensity and income inequality [41,42,43,44], while resilience is assessed by indicators such as social cohesion, infrastructure resilience, crisis management capacity and ecosystem self-repair potential [43,45]. The two concepts provide a complementary perspective for achieving sustainable development goals.

The literature shows that the concept of economic resilience is addressed with different indicators, and Hallegatte [46], while defining the conceptual framework of macroeconomic resilience, stated that GDP is one of the indicators used to measure output. Indeed, the use of GDP to measure economic resilience is widespread in the literature [47,48] and for this reason, GDP is preferred as an indicator of economic resilience in this study.

This study examines the impact of tourism on sustainable growth and asks whether it creates a resilience effect in the economy. The literature on this topic focuses on the tourism-led growth (TLG) hypothesis. The TLG hypothesis is regarded as an extension of the export-led growth hypothesis, given the similarities in the benefits it offers to a country, particularly in terms of economic growth [13]. The triggering of growth is contingent upon the utilization of the obtained foreign exchange in industrial investments, the expansion of production due to the increase in domestic demand, and the enhancement of productivity in production factors. Furthermore, this process provides sectoral diversification and increases employment. Conversely, the non-economic benefits of tourism may include increased welfare due to increased employment, greater sensitivity to the protection of cultural heritage and the environment, enhanced integration with the global community, and even improvements in country governance [13,49,50,51,52,53,54]. In this context, the impact of tourism revenues on the sustainable economy and country development is a topic of interest for policymakers.

Brida et al. [13] and Alcalá-Ordóñez et al. [49], in their comprehensive literature review studies for consecutive periods, support the tourism-led growth view by stating that empirical papers find a positive relationship between tourism and economic growth. Some studies have also identified a reciprocal relationship between economic growth and the tourism sector. Nevertheless, it is emphasized that this interdependence, particularly the influence of tourism revenues on economic growth, is more pronounced in small or middle-income economies with a robust tourism sector [52,55,56,57,58,59]. At this point, drivers that affect tourism performance come to the fore and countries focus on increasing their competitiveness through relevant indicators [60,61,62]. While the factors affecting tourism performance are variable, the contribution of tourism to economic growth cannot be independent of the characteristics of countries. For example, Watson and Deller [63], in their article examining tourism-related economic resilience in the US, found that regions with a high share of tourism in the economy have low resilience due to the impact of tourism volatility on the economy, while in other countries, tourism may be a component that increases economic resilience.

Brida et al. [13] and Alcalá-Ordóñez et al. [49] report the use of several econometric methods in this regard. Although there are many studies that treat the relationship symmetrically, there are a limited number of studies on the asymmetric effect. Wu et al. [64] report that the importance of tourism to regional economies in China has increased over time. The study analyzed eight tourism regions with asymmetric effects, and the positive effect of a positive shock to tourism incomes on growth was observed in only three regions, while the relationship between negative shocks was found in only one region. The study emphasizes the importance of regional differences and argues that this is due to the differentiation of policies implemented, which in turn affects the relationship between tourism and economic growth.

Although tourism is expected to make a positive contribution to Turkey’s economic growth in the long term, the wide variation in the results of studies on Turkey supports further studies on this issue [65]. Furthermore, based on the literature discussed above, the relationship between tourism and growth has mostly been treated symmetrically and in contrast to these studies, we analyze the relationship between tourism and growth asymmetrically [66,67,68]. A review of the literature reveals no publications that address TLG in detail for a single country (Turkey) in the context of resilience with an asymmetric perspective. This is the novelty of the current study.

3. Data, Method and Hypotheses

It is true that the symmetrical relationship between economic growth and tourism income is widely examined in the existing literature. However, in this study, we use the Granger–Yoon methodology to reveal the asymmetric relationships between the variables in question. This innovative approach differs from the aforementioned literature by providing further insights into how positive and negative developments in the tourism sector can explain corresponding positive and negative developments in economic growth. The starting point for our time period is 1990. There are two specific reasons why this date is the reference point for our study. First, while significant developments in the tourism sector began in 1982, the actual realization of tourism activities became evident in 1990. Second, relevant literature [69,70,71] in Turkey also uses this date as a reference point. On the other hand, the availability of data limits the period to 2023. We obtained the data on GDP per capita (current USD)—GDP series, spanning from 1990 to 2023, from the World Bank online database. Simultaneously, we sourced the net tourism income (current USD) series—TOI, covering the same time frame, from the Association of Turkish Travel Agencies online database.

Initially, we conducted unit root tests to assess the stationarity of both series. To this end, we utilized the Phillips–Perron test (PP) developed by Phillips and Perron [72] for examining unit root presence. In addition, we examined the unit root in the presence of structural breaks within the series using the Zivot and Andrews unit root test (ZA) [73], which allows for a single sharp structural break, and the Fourier-ADF unit root test (F-ADF) of Enders and Lee [74], which allows for gradual structural breaks.

To uncover short-term asymmetric causality between the series, we employed the Hatemi-J [75] test (HJ). According to the test’s methodology, we were able to express the series as Equations (1) and (2), given that the series are integrated of order 1 (I(1)).

$${GDP}_t={GDP}_{t-1}+{\epsilon }_{1t}={GDP}_0+{\sum }_{i=1}^t{\epsilon }_{1i}$$

(1)

$${TOI}_t={TOI}_{t-1}+{\epsilon }_{2t}={TOI}_0+{\sum }_{i=1}^t{\epsilon }_{2i}$$

(2)

In Equations (1) and (2), ${GDP}_0$ and ${TOI}_0$ represent the initial values, while the series ${\epsilon }_{1t}$ and ${\epsilon }_{2t}$ denote white noise. The positive and negative shocks of the series are derived as ${\epsilon }_{1i}^{+}=max\left({\epsilon }_{1i},0\right)$, ${\epsilon }_{2i}^{+}=max\left({\epsilon }_{2i},0\right)$, ${\epsilon }_{1i}^{-}=min\left({\epsilon }_{1i},0\right)$, ${\epsilon }_{2i}^{-}=min\left({\epsilon }_{2i},0\right)$ [76]. By using this breakdown, we can express that ${\epsilon }_{1i}={\epsilon }_{1i}^{+}+{\epsilon }_{1i}^{-}$ and ${\epsilon }_{2i}={\epsilon }_{2i}^{+}+{\epsilon }_{2i}^{-}$. The cumulative positive and negative shocks of the series can be obtained as shown in Equations (3)–(6).

$${GDP}_t^{+}={\sum }_{i=1}^t{\epsilon }_{1i}^{+}$$

(3)

$${GDP}_t^{-}={\sum }_{i=1}^t{\epsilon }_{1i}^{-}$$

(4)

$${TOI}_t^{+}={\sum }_{i=1}^t{\epsilon }_{2i}^{+}$$

(5)

$${TOI}_t^{-}={\sum }_{i=1}^t{\epsilon }_{2i}^{-}$$

(6)

In this phase, the HJ test procedure involves treating the cumulative shocks of each series as a VAR model with p lags. The lag p represents a lag for which a VAR model satisfies all stability conditions, and it is determined using information criteria like AIC or SC. The HJ test employs the Wald test statistic (W-stat.) to examine asymmetric causality between the cumulative shocks, with the utilization of bootstrap critical values [75].

To uncover long-term asymmetric relationships between the series, we conducted hidden co-integration techniques as outlined by Granger and Yoon [15], based on the co-integration approach by Engle and Granger [77]. The procedure of hidden co-integration employed the cumulative shocks from Equations (3) to (6) as variables in both the long-term and short-term equations.

Table 2. Definitions of asymmetric tourism-led growth hypotheses.

Case	Sign *	Shocks of TOI	Shocks of GDP	Hypothesis
1	$+$	${TI}^{+}$	${GDP}^{+}$	Enhancing effect
2	$-$	${TI}^{-}$	${GDP}^{+}$	Inhibiting effect
3	$-$	${TI}^{+}$	${GDP}^{-}$	Parachute effect
4	$+$	${TI}^{-}$	${GDP}^{-}$	Exacerbating effect

*: Sign of the coefficient of the shocks of TOI variable in long-run equation.

outlines the definitions of the asymmetric tourism-led growth hypotheses. In this table, the shocks of TOI are considered independent variables, while the shocks of GDP are treated as dependent variables for the long-term equations. Additionally, the coefficients for the shocks of TOI in the equations are assumed to be statistically significant. Consequently, the procedure results in the derivation of four distinct equations to test the asymmetric hypotheses.

Generally, the standard (symmetric) tourism-led growth hypothesis suggests that increasing tourism incomes result in increased economic growth. With regard to the scenarios outlined in Case 1 in Table 2, a positive shock to tourism incomes would lead to an increase in positive shocks for GDP. We termed this effect the “enhancing effect”. Consequently, the asymmetric tourism-led growth hypothesis is applicable in this case. Case 2 indicates that a negative shock to tourism income would lead to a decrease in economic growth, implying that negative shocks in tourism serve to mitigate economic growth. This effect was designated the “inhibiting effect”. Hence, the asymmetric tourism-led growth hypothesis holds true for Case 2. However, in Case 3, a positive shock to tourism incomes triggers a reduction in economic decline, signifying that positive shocks in tourism curb economic downturns and we termed this effect the “parachute effect”. Therefore, the asymmetric tourism-led growth hypothesis is applicable to Case 3. Lastly, in Case 4, a negative shock to tourism causes an increase in economic decline, suggesting that negative shocks in tourism incomes act as an impulsive force for economic downturn. This effect was designated the “exacerbating effect”. Consequently, there exists evidence supporting the validation of the asymmetric tourism-led growth hypothesis.

4. Empirical Findings

Firstly, the stationarity of the series was examined using the PP test from ordinary unit root tests. The results are given in

Table 3. The results of unit root tests.

PP Test	Adj. t-stat	Critical Value
$GDP$	−1.781	−4.27
$∆GDP$	−5.761 *	−4.28
$TOI$	−2.584	−4.27
$∆TOI$	−14.836 *	−4.28
ZA Test	t-stat	${\mathit{T}}_{\mathit{b}}$	Lag	Critical Value
$GDP$	−3.668	2004	0	−5.57
$TOI$	−4.388	2008	0	−5.57
F-ADF Test	Lag	$\widehat{\mathit{k}}$	$\mathit{F}\left(\widehat{\mathit{k}}\right)$	Critical Value	${\mathit{\tau }}_{\mathit{D}\mathit{F}\_\mathit{t}}$
$GDP$	0	1	3.848	12.21	−2.627
$TOI$	0	1	6.162	12.21	−4.371

The models with trend and intercept are used in all tests. The critical values at 0.01 level are presented. *: Significant at 0.01 level.

. According to the PP test results, the TOI and GDP series are non-stationary, and stationary in the first difference at the 0.01 significance level.

Non-stationarity in a time series can be caused by structural breaks. Therefore, the ZA unit root test, which allows for a sharp structural break, and the F-ADF test, which allows for smooth transition structural breaks, are used to overcome this problem. The results of both tests are presented in Table 3. According to the results of the ZA test, although a break is observed in the GDP series in 2004 and in the TOI series in 2011, the unit root hypothesis cannot be rejected. Therefore, these breaks are insignificant, and the series are not stationary.

In the ZA test, lag values are determined by AIC statistic. The results of the F-ADF test are also provided in Table 3. The Fourier augmented Dickey–Fuller (F-ADF) equation, as suggested by Enders and Lee [74], was estimated for $1\le k\le 5$ values, and the frequency $\widehat{k}$ for both series corresponding to the equation with the smallest sum of squared residuals (SSR) was determined as 1. The significance of the Fourier terms in the equation was tested using the Wald test, and according to the $F\left(\widehat{k}\right)$ statistic, it was observed that these terms were not significant at the 1% level. The critical values were obtained from Enders and Lee [74]. Consequently, there are no smooth transition breaks in either series. Therefore, there was no need for the suggested test statistic ${\mathsf{\tau }}_{\mathrm{D}\mathrm{F}\_\mathrm{t}}$ for testing the unit root hypotheses.

According to all the unit root test results conducted, it was concluded that both the analyzed TOI and GDP series are integrated of order 1 (I(1)).

To obtain short-run asymmetric causalities, the HJ test can be conducted. This involves obtaining the cumulative shocks of the series as per Equations (3)–(6). The lag value is determined from the VAR model, which satisfies all stability conditions based on the AIC statistic. The results of the HJ test are displayed in

Table 4. Short run asymmetric causalities.

		Bootstrap Critical Values (10,000 Replications)
HJ Test	Wald Stat.	0.01	0.05	0.10
${TOI}^{+}\ne >{GDP}^{+}$	90.728 **	115.198	39.177	22.361
${TOI}^{-}\ne >{GDP}^{+}$	0.852	10.522	5.288	3.422
${TOI}^{+}\ne >{GDP}^{-}$	4.617 *	12.760	5.605	3.608
${TOI}^{-}\ne >{GDP}^{-}$	110.986 **	32.147	11.797	7.433
${GDP}^{+}\ne >{TOI}^{+}$	14.184	81.277	25.765	15.238
${GDP}^{-}\ne >{TOI}^{+}$	0.380	87.651	26.493	16.010
${GDP}^{+}\ne >{TOI}^{-}$	5.334	92.034	29.228	17.405
${GDP}^{-}\ne >{TOI}^{-}$	0.020	24.876	10.007	6.529

$\ne >$: No asymmetric causality. *: significant at 0.10 level, **: significant at 0.05 level.

.

Based on the findings presented in Table 4, there exists an asymmetric causality from positive shocks in tourism incomes to positive shocks in GDP in the short term. Similarly, negative shocks in tourism incomes cause negative shocks in GDP in the short term.

As can be seen, the causality runs from tourism to economic growth and the interaction suggests a causality between tourism and economic growth in the short run. In other words, short-term policies aimed at fostering tourism may directly impact the economy, either positively or negatively, in Turkey. Furthermore, the results show an asymmetric causality from positive shocks to tourism income to negative shocks to GDP in the short term.

Although the causality analysis shows that there is an interaction in the short run, it does not provide information on whether the interaction is positive or not, i.e., whether tourism makes a positive contribution to the economy or not. In order to provide more insight for the formulation of effective long-term policies, the long-term relationships between the variables are examined. To establish long-run asymmetric relationships between the series, we initially conducted the Granger and Yoon [15] hidden co-integration test (GY) based on the co-integration approach by Engle and Granger [77]. Prior to this test, unit root tests were performed for the positive and negative cumulative shock series, as defined in Equations (3)–(6), confirming that all cumulative shocks were integrated of order 1 (I(1)). These unit root test results are not included here to conserve space. The GY test was conducted for the functions ${GDP}_t^{+}=f\left({TOI}_t^{+}\right)$, ${GDP}_t^{+}=f\left({TOI}_t^{-}\right)$, ${GDP}_t^{-}=f\left({TOI}_t^{+}\right)$ and ${GDP}_t^{-}=f\left({TOI}_t^{-}\right)$ using a model with the intercept. According to the findings, the GY test revealed co-integration only for the function ${GDP}_t^{-}=f\left({TOI}_t^{+}\right)$. This result is detailed in

Table 5. Hidden co-integration results.

Null: No Co-Integration
tau Stat.	p-Value	z Stat	p-Value
−2.99	0.139	−16.36 *	0.063
Long run equation
$\mathrm{Dep}. \mathrm{Var}.: {GDP}_t^{-}$	Coeff.	HAC St. Error	t Stat	p-Value
Cons.	−0.0835	0.206	−0.40	0.687
${TOI}_t^{+}$	−0.2932 **	0.123	−2.37	0.023
CECM

$∆{GDP}_t^{-}=-0.064+0.245∆{TOI}_{t-3}^{+}, {\overline{R}}^2=0.11$. $∆{TOI}_t^{+}=-1.047{\epsilon }_{t-1}+0.325∆{TOI}_{t-1}^{+}+0.631∆{TOI}_{t-4}^{+}$, ${\overline{R}}^2=0.25$. *: Significant at 0.10 level, **: significant at 0.05 level.

.

Based on the z statistics provided in Table 5, it is evident that the series ${GDP}_t^{-}$ as the dependent variable and ${TOI}_t^{+}$ are co-integrated at the 10% level. Detailed insights into this long-run estimation are available in Table 5. Notably, the residuals from this estimation demonstrate a normal distribution, as indicated by Jarque–Bera test (JB = 1.266 and p-value = 0.531). To account for potential heteroscedasticity and autocorrelation in residuals, a correction was made through the calculation of the HAC variance–covariance matrix. A thorough examination of the long-run estimation results indicates the statistical significance of the regression coefficient at 5% level (p = 0.023 < 0.05). Particularly noteworthy is the negative coefficient associated with ${TOI}_t^{+}$, corresponding to Case 3 outlined in Table 2. This finding suggests that an increase in positive shocks related to tourism incomes will result in a decrease in negative shocks on GDP in the long run, namely, a 1% increase in positive shocks to tourism income leads to a 0.29% decrease in negative shocks to GDP. In simpler terms, while the increase in tourism income makes no contribution during periods of economic growth, it is observed that it reduces the level of contraction during periods of economic contraction, thus realizing the parachute effect defined in Table 2.

These results partially support the validity of the asymmetric tourism growth hypothesis in the context of Turkey. Consequently, these results confirm that tourism revenues play a role in Turkey’s sustainable growth objectives. It is recommended that policy makers, especially those focusing on the tourism sector, develop strategies to stabilize and, if possible, increase tourism income in Turkey.

To obtain Equations (7) and (8) for estimating CECM, refer to the work of Granger and Yoon [15].

$$∆{GDP}_t^{-}={\psi }_0+{\psi }_1{\widehat{\epsilon }}_{t-1}+{\sum }_{i=1}^k{\psi }_i\mathsf{\Delta }{TOI}_{t-i}^{+}+{\sum }_{j=1}^p{\psi }_j\mathsf{\Delta }{GDP}_{t-j}^{-}+{\upsilon }_t$$

(7)

$$\mathsf{\Delta }{TOI}_t^{+}={\gamma }_0+{\gamma }_1{\widehat{\epsilon }}_{t-1}+{\sum }_{i=1}^k{\gamma }_i\mathsf{\Delta }{TOI}_{t-i}^{+}+{\sum }_{j=1}^p{\gamma }_j\mathsf{\Delta }{GDP}_{t-j}^{-}+v_t$$

(8)

In Equations (7) and (8), ${\widehat{\epsilon }}_{t-1}$ represents the one-lagged residuals of the long-run equation estimated in Table 5. The coefficients ${\psi }_1$ and ${\gamma }_1$ indicate the long-run adjustments and the short-run adjustments are represented by the coefficients of the lagged difference variables. Previous studies, such as those by Granger and Yoon [15] and Honarvar [78], have reported only significant coefficients in these CECM equations. In this study, a stepwise regression proposed for CECM by Mert and Caglar [76,79] was conducted to obtain equations with significant coefficients. The maximum lag value (k and p) is set at 5, and stepwise regression with backward selection is applied at a significance level of 0.05 to derive CECM equations with significant coefficients at the 0.05 level. The CECM estimation results are also presented in Table 5.

In the short-term equations as CECM provided in Table 5, all coefficients are significant at the 0.05 level due to the stepwise regression procedure. According to Gonzalo and Granger [80], Mert and Caglar [76,79], the CECM equation’s dependent variable with an insignificant error correction coefficient represents the permanent component, responsible for the long-term dynamics between variables. Any shock to this permanent component affects itself and the other variable in the long run. Conversely, the dependent variable of the CECM equation with a significant error correction coefficient is defined as transitory. A transitory variable does not determine long-term dynamics, and any shock to it is impermanent in the long run.

Based on the results of CECM in Table 5, the error correction coefficient is insignificant for the short-term equation of ${GDP}_t^{-}$ (negative shocks of GDP), while the error correction coefficient is negative and significant (coef. = −1.047, p < 0.05) as anticipated for the short-term equation of ${TOI}_t^{+}$. Thus, the series ${GDP}_t^{-}$ represents the permanent component, whereas the series ${TOI}_t^{+}$ is transitory in the long run. Negative shocks to GDP determine the long-term dynamics, which are the cause of the positive shocks to tourism income. In other words, although the economic contraction does not have a direct long-term effect on tourism, it is observed that it has an asymmetric effect within the framework of the error correction model.

5. Discussion

In this study, we examine the asymmetric tourism-led growth hypotheses in Turkey using data from 1990 to 2023. There are many studies in the literature that assess the relationship between tourism and economic growth, mostly under the concept of the tourism-led growth hypothesis. Most studies assume that the relationship between tourism and economic growth is symmetric and test this relationship with panel data including many countries. In this literature, the hypothesis is confirmed for some countries and not for others [81,82,83,84,85].

Turkey is a country whose economy and tourism frequently face internal and external shocks (Table 1). The tourism sector contributed to the GDP of USD 1.024 trillion in 2023 by generating USD 54.3 billion in revenue [25,86]. In other words, the income generated is 5.3 per cent of GDP. Although this rate may seem low compared to the European average of 7.2%, the inflow of foreign exchange generated by tourism has the potential to provide resilience to the Turkish economy, which often faces the problem of a current account deficit [87]. As stated in the article, resilience is the ability of systems to survive, adapt and recover from shocks and changes. In order to ascertain the resilience effect, it is imperative to undertake a detailed examination of the relationships that occur during periods of shock. In this regard, asymmetric models have the capacity to furnish more accurate policy recommendations by unveiling the differentiated effects that are observed during periods of positive and negative changes in variables [88,89,90,91,92].

In this study, in order to capture the detailed impact of tourism on growth during periods of shock, we examine short- and long-run asymmetric relationships between positive and negative shocks to tourism receipts and GDP. We then apply hidden cointegration techniques to the series and estimate crouching error correction models.

Based on the results of short-run causality, there is unidirectional asymmetric causality from the positive shocks of tourism incomes to the positive shocks of GDP $({TOI}^{+}=>{GDP}^{+})$, from the negative shocks of tourism incomes to the negative shocks of GDP $({TOI}^{-}=>{GDP}^{-})$, and also from the positive shocks of tourism incomes to the negative shocks of GDP $({TOI}^{+}=>{GDP}^{-})$. These results suggest that in all cases there is an interaction between tourism and economic growth in Turkey in the short run. In other words, there is a sign related to the TLG hypothesis. The relationships between positive tourism shocks and economic growth are similar to Eyuboglu and Eyuboglu [67]. Ideally, these relationships should also be achieved in the long term.

When using hidden cointegration analysis for the long run view, the negative coefficient of ${TOI}^{+}$ in the long run equations (Table 5) shows that the series and ${TOI}^{+}$ and ${GDP}^{-}$ are cointegrated, meaning that the positive trends in tourism receipts have a significant dampening effect on the downward movements of GDP. We refer to this effect of slowing down the economic decline as the parachute effect and, similar to Kumar et al. [68], this effect is seen as a sign of resilience that is reflected in the overall economy. Indeed, our finding is supported by studies showing that tourism has a high capacity to generate economic resilience [63,93].

The results of the crouching error correction model in detail reveal that ${GDP}^{-}$ represents the permanent component, while ${TOI}^{+}$ acts as the transitory component in the long run. Therefore, the fluctuations in the economy are accountable for the long-term dynamics, and a negative shock to the economy in Turkey will impact both an increase in tourism income and the economic contraction itself.

These results show that in the short term there is an interaction between tourism and economic growth, but in the long term, this relationship remains at the level of a resilience effect. In fact, tourism has outperformed overall growth in all years of economic crises in Turkey. In the 1994, 2001 and 2018–2019 periods, which are characterized as economic crises, tourism provided resilience to the economy by recording growth of 9%, 37% and 13–27%, respectively (Figure 1).

One of the main factors behind the growth of tourism in these periods was the depreciation of the Turkish Lira against other currencies, which increased the price competitiveness of Turkish tourism in these periods. Therefore, tourism acts as a parachute to slow down the decline in periods when Turkey has a foreign exchange deficit. Shahzad et al. [94] reached a similar conclusion by stating that tourism is a driving factor in Turkey during economic downturns. However, the studies that have tested the TLG for Turkey have reached divergent conclusions, with some detecting a relationship through different methods [95,96,97] and others failing to do so [98,99]

With a share of 5.4% of GDP, it is very difficult for tourism to support growth in Turkey with a long-term, continuous and definite impact. We believe that it is possible to increase this impact by taking structural steps under three main headings. The first is to increase the added value of the tourism sector. This is largely based on market and product diversification. Currently, the locomotive of Turkish tourism is Russian and German tourists, and the sector suffers in cases of any unfavorable situations with Russia or Germany. Increasing the number of markets targeted will reduce the frequency and impact of negative situations and increase the added value of a product [100]. Agazade [101] found that international tourism receipts increased and volatility decreased in Turkey when source country diversification was adopted.

As for product diversification, it means opening up areas such as culture, sport, health and congress tourism in addition to sea, sand and sun tourism, and targeting tourists with different income levels and socio-cultural backgrounds. Turkey has great opportunities in adventure, eco and nature tourism with its natural and geographical diversity; sports, congress and fair tourism with the quality of its facilities and accommodation; experiential tourism with its rich culture and gastronomy; health tourism with its health and accommodation infrastructure; and cultural and religious tourism with its historical richness. Taking the right steps, especially with a focus on rural tourism, will increase the prosperity of rural areas. For example, Romao [102] emphasizes the need for product diversification in destinations with high tourism intensity and suggests rural tourism as a strong option. In other words, tourism has the opportunity to generate much more added value to the economy than it creates itself through the multiplier effect [3,4,5,6].

The second is to take measures to prevent fluctuations in the economy and in the tourism sector. Indeed, the CECM results in our study support this recommendation. This heading is quite comprehensive and basically covers both economic and political measures. Steps regarding the economy in general and tourism policy should be taken with a long-term perspective. A stable exchange rate will increase predictability and reduce volatility, both in the overall economy and in the tourism sector. At this point, it is important to implement policies that keep under control factors such as inflation and the current account deficit, which are direct determinants of the exchange rate. It is also important that tourism infrastructure and investments are implemented through transparent planning shared with stakeholders, rather than fluctuating in an unstable manner. Finally, the main determinant of tourism is the security and peace policies developed with neighboring countries. In conclusion, we believe that the success to be achieved by these policies will serve to protect the entire economy from fluctuations along with tourism.

The third is the implementation of policies that respect the principles of environmental, economic and social sustainability. Tourism is a sector with strong links to sustainability principles. UNEP and WTO [103] state that the tourism sector has a relatively higher potential to contribute to social development than other sectors, and that the active participation of local decision-makers in tourism planning is important. It does not seem possible to create sustainable value in the tourism sector while destroying the environment and social fabric. Indeed, as the authors of [104] point out, in order to ensure sustainability in the tourism sector, it is necessary to maintain a balance between economic development and resource conservation.

6. Conclusions

This study aims to observe and evaluate the relationship between tourism and economic growth for Turkey from an asymmetric perspective, which is rare in the literature. There are a large number of publications that address this relationship with a symmetric approach within the framework of the TLG hypothesis. In this study, the asymmetric approach is preferred, which treats the effects of positive and negative shocks separately and provides the opportunity to see the hidden relationship that cannot be detected with the symmetric approach. Turkey, with the crises it has experienced since 1990, is a good example to question the differentiated relationship between tourism and economic growth in periods of positive and negative shocks (Table 1).

First, the Hatemi-J asymmetric causality test is used to test for short-run asymmetric causality, and unidirectional causality is observed between increasing tourism incomes and economic growth, and between decreasing tourism incomes and economic contraction.

Subsequently, the hidden co-integration test revealed a valuable result that tourism slowed down the decline of the Turkish economy like a parachute during the downturn. In other words, tourism provides resilience to the economy. To the best of our knowledge, this is the first study that reveals the resilience effect of tourism in the Turkish economy.

In this study, the issue was only addressed in Turkey, and it is quite possible that different results will be obtained in countries with different characteristics, and comparing the results will allow new conclusions to be drawn. We also believe that the contribution of an asymmetric perspective in tourism will pave the way for new implications for other sectors. Therefore, in future studies, the questioning of growth and sustainability indicators with an asymmetric perspective will reveal the nature of the relationship in more detail. Finally, the use of other variables such as the number of tourists instead of tourism receipts and other development indicators instead of economic growth in subsequent studies will enrich the knowledge in the field.