The Influence of Heritage on the Revealed Comparative Advantage of Tourism—A Worldwide Analysis from 2011 to 2022

Abstract

A country’s development is crucially determined by its cultural and natural heritage, and it is reflected in its industrial structure and its success in the global marketplace. The present paper looks at the global performance of tourism, comparing its performance measured by the Normalised Revealed Comparative Advantage (NRCA) index to the components of natural and cultural heritage, analysing 117 countries of the world. Natural and cultural heritage indicators were derived from the tourism competitiveness reports of the World Economic Forum for the years 2011–2013–2015–2017–2019–2022. Panel regression analysis was applied, with NRCA as the dependent variable and eight indicators of natural and cultural heritage as independent variables, comparing regions of the world. The main findings show considerably differing patterns between regions; Europe and Eurasia being similar to the Americas, with decreasing competitive advantage associated with more focus on endangered species and observance of environmental treaties, while the Middle East and North Africa show a strongly opposite pattern. Cultural heritage has a positive impact only in Sub-Saharan Africa, while Asia and the Pacific benefit mainly from the increase of protected areas and abundance of species. These differences shed light on differences in tourism competitiveness in the global market and may guide policymakers towards utilising heritage items for improving tourism performance.

1. Introduction

A country’s development is crucially determined by its heritage, including its natural and cultural resources, history, and traditions, often through investment and job creation [1], the recognition and development of local identity [2,3], or by driving learning and innovation [4,5] and stimulating economic diversification [5,6]. Natural heritage activities contribute to rural development, especially through environmental management, primary production, and providing a background to other business activities [7]. In the face of climate change, promoting the process of a circular economy, it has become more and more obvious that the acknowledgement of past traditions and the protection of natural and cultural resources are main components of sustainable development, as a demonstration of the unique potential of heritage [8].

The Sustainable Development Goals (SDGs) also underline the fundamental importance of heritage, as SDG11.4 focuses on the protection and safeguarding of the world’s cultural and natural heritage [9]. Similarly, the 2015 UNESCO Policy on World Heritage and Sustainable Development acknowledges the fundamental importance of protecting heritage, with the need of strengthening the three dimensions of sustainable development, environmental sustainability, inclusive social development, and inclusive economic development [9] in the global marketplace, reflecting its capabilities of producing a commodity that is important for the rest of the world. The most successful industries of a country are usually those for which the natural, physical, and human resources are most suited [10,11,12]. Thus, the existence and appreciation of traditions and the protection of natural and cultural heritage are important factors of socioeconomic development.

It is obvious that various industries depend on heritage factors to a varying extent. The tourism industry is one example in which success in the global market crucially depends on natural and cultural heritage. This is reflected by the Travel and Tourism Competitiveness Reports [13,14] and the recent modifications of these in the Travel and Tourism Development reports of the World Economic Forum [15,16]. These reports list the most important heritage items that are relevant for the tourism industry and provide quantitative indicators, defining the complex Travel and Tourism Competitiveness Index (TTCI) for most countries of the world.

The present paper looks at the global market success of the travel and tourism industry in contrast to the natural and cultural heritage of 117 countries of the world. Indicators of various items of natural and cultural heritage were derived from the World Economic Forum (WEF) databases for the years 2011–2013–2015–2017–2019–2022 [17,18,19,20,21,22,23,24]. The market success of the tourism industry is measured by a modified version of the Revealed Comparative Advantage Index, RCA [25], which was developed in 2009 under the name of Normalised Revealed Comparative Advantage, NRCA [26], as an improvement eliminating the known weaknesses of the original RCA index [26,27,28].

The components of the TTCI—including cultural and natural heritage indicators—are applied to assess the resources and endowments of countries as facilitators of tourism development, e.g., and their impact on international tourism arrivals and receipts [29,30,31,32].

The RCA and the NRCA indices have already been used for several regions of the world to assess the market success of the tourism industry globally and for specific geographic regions, e.g., Eastern Africa [33], Sub-Saharan Africa [34], and the continents of the world [31].

The novel aspect of our research, however, is to pair the two approaches, i.e., revealed comparative advantage as a measure of the market performance of the tourism industry and heritage indicators as major resources. This approach, though applied for specific regions of the world, has not been pursued on this large a scale, comparing 117 countries covering the main geographical regions, nor on this timescale, from 2011 to 2022, using the latest available data on heritage values. Thus, the analysis can give a good global insight into the contribution of heritage to the export competitiveness of tourism, both reflecting temporal changes and regional differences of the world.

2. Materials and Methods

2.1. Data

The analysis covers 117 countries of the world, in the following structure, with Africa having 25 countries in the analysis, the Americas 21, Asia and the Pacific 26, Europa 37, and the Middle East eight. The choice of the countries was determined by the availability of data regarding tourism performance, total exports, and data on cultural and natural heritages available for the 2011–2022 time period. See

Table 1. Variables and data sources.

Variable Notation	Variable Name	Meaning	Data Source
WHSC_no	Number of world heritage cultural sites	Number of world heritage cultural sites in country	TTCI/TTDI datasets [17,18,19,20,21] and reports [15,16,22,23,24] by WEF
WHSN_no	Number of world heritage natural sites	Number of world heritage natural sites in country
KnownSpecies	Total known species	Total known species in the country
EnvTreaty	Environmental treaty ratification	Number of environmental treaties ratified by country
TServInf	Tourist service infrastructure	Scale of 1 (very bad) to 7 (very good)
ProtAreaPct	Protected area percent	Total protected area, as % of country area	[35]
ForestPct	Forest area percentage	Total forest area, as % of country area	[36]
RedListInd	Red List Index	Aggregate extinction risk (0 to 1); with 0: all species extinct; 1: no species to be extinct in near future	[37]
RCA	Revealed Comparative Advantage Index for tourism	Computed by Equation (1) for tourism	International tourism receipts: [38,39]; Total export values: [40]
NRCA	Normalised Revealed Comparative Advantage Index for tourism	Computed by Equation (2) for tourism

for the data sources.

The list of the countries is presented in Appendix A. The analysed variables are the NRCA index, computed by the author based on the international tourism revenues and the total export revenues of the countries, compared to the world total and the availability of various heritage items—components of cultural and natural heritage—as listed by the Travel and Tourism Competitiveness Reports and Travel and Tourism Development Reports by the WEF [14,15,16,22,23,24,25]. The heritage indicators are accompanied by the quality of the tourism service infrastructure, as this feature certainly influences the accessibility of heritage for tourism purposes. The analysis purposefully avoids the inclusion of general economic variables, such as GDP or price levels, as these might mask the influence of heritage on export competitiveness. The complete list of the variables, with their precise meanings and sources of data are listed in Table 1. Some variables were not covered by the WEF reports for all countries or all years, and, in these cases, the research used the original full data sources suggested by the WEF reports. The timescale of the analysis is also limited by the coverage of the WEF reports (6 years: 2011, 2013, 2015, 2017, 2019, and 2022).

2.2. The Computation of the NRCA Index

The index of Revealed Comparative Advantage (RCA) defined by Balassa in 1965 [25] is a tool to assess the export competitiveness of countries. It is computed by the following formula, with values between 0 and 1 meaning comparative disadvantage, while values above 1 reflect comparative advantage (see Equation (1)):

RCA_c,i = [X_c,i/X_c]/[X_w,i/X_w] = [X_c,i/X_w,i]/[X_c/X_w],

(1)

where X: export value, c: 1…m country, i: 1…n industry, w: world, so X_c = ∑ _(i=1…n) X_c,i and, similarly, X_w,i = ∑ _(c=1…m) X_c,i and X_w = ∑ _(c=1…m) X_c.

The Revealed Comparative Advantage index (RCA) compares the share of a country’s total exports of the analysed commodity in its total exports to the share of world exports of the same commodity in total world exports. An index value above 1 means that the analysed commodity represents a higher share in the country’s total exports than in the world, and this indicates that the country has an advantage in the commodity compared to the world average.

Although RCA is widely used in trade competitiveness analysis, there are some problematic features that may distort its meaning [26,27,28]. The two most important problems are as follows: (a) RCA is not symmetric because non-competitive values fall between 0 and 1, while competitive values range from 1 to infinity; (b) the so-called small-country bias means that, when X_c is small relative to X_w, then RCA tends to be high, suggesting unreasonably high comparative advantage. To correct these weaknesses, the index of Normalised Revealed Comparative Advantage (NRCA) was developed by Yu et al. [26], as in Equation (2):

NRCA_c,i = X_c,i/X_w − X_c × X_w,i/X_w² = [X_c,i/X_w,i − X_c/X_w] × X_w,i/X_w

(2)

with the same notations as the ones used for RCA.

The Normalised Revealed Comparative Advantage index (NRCA) calculates the difference between a country’s actual exports share of world exports in the analysed commodity and the country’s total export share in total world exports. This difference is normalised by the share of the analysed commodity in world exports. If the country’s export share of the commodity is the same as its total export share, then the commodity is in a neutral position in the country, producing the same export share as the average commodities of the country. In this case, NRCA equals zero, while positive values represent a successful export commodity, and negative values indicate an unsuccessful one disadvantaged in the world market.

As Yu et al. [26] demonstrated, the NRCA is symmetric, ranging from −1 to +1, with the neutral value being zero, and it is free of small-country bias. In addition, NRCA is additive and transitive, and it correctly assesses temporal changes, i.e., it allows for between-year comparisons.

The application of NRCA for trade analysis is justified by its ability to handle the weaknesses of RCA, as is explained in the Materials and Methods section. As was stated in [27], NRCA has the unique ability to be additive with respect to both countries and commodities, and it is consistent with regard to time. This feature makes NRCA values comparable between countries, commodities, and time periods.

Despite the problems of the original RCA index, it has been applied for the tourism industry, using international tourism revenues as the indicator of tourism exports at the country level [33,34,41]. However, Bogale [34] also used NRCA for tourism, revealing important differences between the assessments by the two methods.

2.3. Analysis of the Relationship between NRCA and Heritage

The NRCA values are compared to the heritage indicators for the 117 countries. The comparison is performed by correlation analysis, as an initial step, followed by a panel regression for the 117 countries and the 6 years, with NRCA being the dependent variables, and heritage indicators and tourism service infrastructure being the independent ones. Panel regression was used, focusing on fixed effects of the independent variables, as the 6 years were not sufficient for doing a proper temporal assessment for the eight independent variables.

The panel regression estimation was performed by the Linear Mixed Model method (LMM) available in the “R” Statistical Software [42]. The assumptions for the LMM require that the explanatory variables are related linearly to the response variable, and the errors have constant variance, are independent, and are normally distributed. These diagnostics are presented as a part of the analysis. However, as Schielzeth et al. [43] proved, the LMM and other mixed-effect procedures are considerably robust, i.e., minor violations of the assumptions do not make the model estimates unreliable.

The general equation of a panel model is [44]

y(i,j) = a + c(i)+ ∑_{k = 1…K} [b(k) + d(i,k)] ×(i,j,k) + e(i,j)

where:

• y(i,j): is the observed dependent variable, with two factors, i = 1…n and j: 1…M (e.g., i being the time and j being the subject or vice versa);
• x(i,j,k): is the value of the observed independent variable k (k = 1…K), for i,j;
• a: is the fixed intercept, not depending on i and j;
• b(k): is the fixed slope of the independent variable k;
• c(i): is the random intercept, varying by factor i;
• d(i,k): is the random slope, varying by variable k and factor i;
• e(i,j): is the random error (of normal distribution).

Panel model estimation requires the estimations of a, c(i), b(k), d(i,k), and e(i,j). Various simplifications exist with regard to assumptions of the parameters, and the estimation methods differ according to these simplifications. Several statistical packages provide estimation procedures. The present research uses the LME4 package [45] of „R” statistical software [42]; therefore, the LME4 formulas are also presented with the equations. In these equations Y and X1, …XK refer to the various variable names. Four model versions are described:

• Fixed intercept and slopes, with random intercepts and slopes: y(i,j) = a + c(i)+ ∑_k=1…K [b(k) + d(i,k)] x(i,j,k) + e(i,j), i.e., the full model. The relevant LME4 formula is Y~1 + X1 + X2 +…XK + (1 + Xk| subject) or Y~1 + X1 +…+ XK + (1 + Xk)|time) depending on which factor and which Xk independent variable are used for random effects.
• Fixed intercept and slope with random intercept: y(i,j) = a + c(i)+ ∑_k=1…K [b(k)] x(i,j,k) + e(i,j). The LME4 formula: Y~1 + X1 +…+ XK + (1|subject) or Y~1 + X1 +…+ XK + (1|time).
• Fixed intercept with random slopes: y(i,j) = a + ∑_k=1…K d(i,k) x(i,j,k) + e(i,j). The LME4 formula: Y~1 + X1 + X2 +…XK + (0 + Xk|subject) or Y~1 + X1 +…+ XK + (0 + Xk|time).
• Fixed slopes and random intercept: y(i,j) = c(i) + ∑_k=1…K b(k) x(i,j,k) + e(i,j). The LME4 formula is Y ~ 0 + X1 + X2 +…XK + (1|subject), or Y ~ 0 + X1 +…+ XK + (1|time).

From the viewpoint of interpretation, Model 4 seems to be the simplest, as it assumes that the average impact of each independent variable on the dependent variable is the same, regardless of subjects or time, but there are variations around this average according to subjects or times. The following analysis applies this relatively simple approach. If the model had turned out to be unreliable, a more complex structure would have to be explored.

3. Results

Table 2. Descriptive statistics of the analysed variables.

	N		Min		Max		Mean	St.Dev.	Region			Year
									K-W St		p	K-W St		p
RCA	595		0.014		12.142		1.968	1.917	13.55820		0.00885	1.53053		0.90952
NRCA	557		−0.305		0.318		0.000	0.049	11.41499		0.02228	2.36870		0.79613
WHSC_no	691		0.000		70		6.943	9.861	105.59840		6.315 × 10−22	4.88886		0.42959
ProtAreaPct	653		0.000		56.5		16.492	11.527	53.76257		5.901 × 10−11	12.12895		0.03306
KnownSpecies	692		0.000		15,878.0		2167.63	2663.90	108.31665		1.663 × 10−22	354.57962		1.807 × 10−74
WHSN_no	688		0.000		16		1.773	2.669	61.31917		1.532 × 10−12	4.31315		0.50527
ForestPct	696		0.000		73.74		29.746	19.688	206.34547		1.623 × 10−43	0.01236		1.00000
Red list index	695		0.000		0.992		0.856	0.115	284.35238		2.568 × 10−60	2.30220		0.80594
TServInf	692		0.000		7.00		3.820	1.410	160.33738		1.238 × 10−33	70.42547		8.358 × 10−14
EnvTreaty	686		0.000		30.00		21.613	3.900	64.62533		3.086 × 10−13	99.37025		7.174 × 10−20
RCA	595		0.014		12.142		1.968	1.917	13.55820		0.00885	1.53053		0.90952
NRCA	557		−0.305		0.318		0.000	0.049	11.41499		0.02228	2.36870		0.79613
Means by region		Sub-Saharan Africa		Europe and Eurasia		Middle East and North Africa				The Americas			Asia and Pacific
WHSC_no		2.65		10.02		5.09				5.52			7.38
ProtAreaPct		15.15		19.22		9.10				18.09			13.87
KnownSpecies		2253.98		1077.09		1145.73				3833.75			3266.27
WHSN_no		1.18		1.43		0.42				2.66			3.01
ForestPct		25.15		30.32		3.85				41.53			36.87
Red list index		0.85		0.93		0.87				0.80			0.76
TServInf		2.7554		4.5878		3.7276				3.8123			3.3432
EnvTreaty		20.675		23.098		19.786				21.032			21.061
RCA		2.14		2.12		2.52				1.65			1.53
NRCA		0.00137		−0.00390		0.00664				0.01098			−0.00978

Note: K-W = Kruskal-Wallis test for significant differences between regions and years.

presents the descriptive statistics of the analysed variables in our database of 2011–2013–2015–2017–2019–2022, for 117 countries, together with regional mean values. The normality testing of the variables rejected the assumption of normality; therefore, we used nonparametric statistical methods that are not sensitive to lack of normalities. The variables were compared for significant differences between regions and years, applying the Kruskal–Wallis nonparametric test. Table 2 shows that significant differences were found between regions for all variables, while, regarding the years, only the ProtAreaPct, KnownSpecies, TServInf, and EnvTreaty variables differed significantly. Figure 1 presents the temporal patterns of each variable by region.

The pairwise correlations (

Table 3. Pairwise correlations of RCA, NRCA, and heritage-related indicators.

		WHSC _no	Prot AreaPct	Known Species	WHSN _no	Forest Pct	RedList Ind	TServ Inf	Env Treaty	RCA
WHSC_no	P	—
	S	—
Prot Area Pct	P	0.148 ***	—
	S	0.156 ***	—
Known Species	P	0.135 ***	−0.016	—
	S	0.139 ***	−0.02	—
WHSN_no	P	0.494 ***	0.087 *	0.454 ***	—
	S	0.471 ***	0.199 ***	0.390 ***	—
ForestPct	P	0.061	0.294 ***	0.240 ***	0.095 *	—
	S	0.069	0.311***	0.229 ***	0.206 ***	—
RedListInd	P	0.017	0.181 ***	−0.284 ***	−0.111 **	−0.119 **	—
	S	0.035	0.129 **	−0.336 ***	−0.168 ***	−0.165 ***	—
TServInf	P	0.293 ***	0.160 ***	−0.237 ***	0.164 ***	0.080 *	0.178 ***	—
	S	0.329 ***	0.167 ***	−0.361 ***	0.106 **	0.076 *	0.135 ***	—
EnvTreaty	P	0.363 ***	0.260 ***	0.184 ***	0.168 ***	0.188 ***	0.231 ***	0.314 ***	—
	S	0.423 ***	0.258 ***	0.228 ***	0.308 ***	0.177 ***	0.103 **	0.281 ***	—
RCA	P	−0.124 **	−0.170 ***	−0.087 *	−0.113 **	−0.065	−0.130 **	0.075	−0.104 *	—
	S	−0.065	−0.119 **	−0.034	−0.04	−0.037	−0.122 **	0.122 **	−0.104 *	—
NRCA	P	−0.091 *	−0.091 *	0.138 **	0.236 ***	−0.058	−0.060	0.114 **	−0.184 ***	0.280 ***
	S	0.044	−0.046	0.099 *	0.059	0.009	−0.186 ***	0.174 ***	−0.096 *	0.816 ***

Significance of correlations: *: p < 0.05, **: p< 0.01, ***: p < 0.001, P: Pearson’s, S: Spearman’s correlation.

) indicate that there are moderate but significant correlations between the NRCA and heritage components. The largest absolute values are with the Red List Index and the Tourism Service Infrastructure variables (Spearman’s correlation). As the data are not normally distributed, the Pearson and Spearman correlations considerably differ. The small absolute values indicate that the pooled data do not show clear patterns, and the various indicators reflect significant relationships between the heritage components and export competitiveness of tourism.

The heritage-related independent variables were tested for collinearity, and, as the TOL and VIF factors indicate (

Table 4. Multicollinearity test for heritage-related variables.

	Tolerance (TOL)	Variance Inflation Factor (VIF)
WHSC_no	0.706	1.416
ProtAreaPct	0.784	1.276
KnownSpecies	0.428	2.338
WHSN_no	0.584	1.712
ForestPct	0.804	1.244
Red list index	0.793	1.261
TServInf	0.649	1.54
EnvTreaty	0.591	1.692
Year	0.512	1.955

), no issues of multicollinearity were detected.

As correlations did not reveal any clear patterns, the analysis continues with a panel regression based on the LME4 package of „R”, as described in the Methodology section. The first step was using all the countries and years for a panel regression with NRCA as a dependent variable, heritage components as independent ones, and year as a random factor. The results are presented in

Table 5. Fixed- and random-effect estimates for all countries.

Variable	Fixed-Effects Estimate	Std. Err	t Value	Pr(>|t|)		Random Intercept for Years
WHSC_no	−0.001205	0.0003	−4.413	1.25 × 10−5	***	2011: −0.0013872238
ProtAreaPct	−0.000217	0.0002	−1.064	0.28775		2013: −0.0064993186
KnownSpecies	0.000003	0.0000	2.447	0.01538	*	2015: 0.0006474683
WHSN_no	0.005909	0.0011	5.462	7.46 × 10−8	***	2017: −0.0010744230
ForestPct	−0.000065	0.0001	−0.487	0.62641		2019: 0.0003912855
RedListInd	−0.022490	0.0216	−1.044	0.2971		2022: 0.0079222115
TServInf	0.009548	0.0020	4.879	1.62 × 10−6	***
EnvTreaty	−0.003565	0.0007	−5.379	1.31 × 10−7	***
Region Sub-Sah Afr	0.061850	0.0211	2.928	0.00358	**
Region Europe & Eurasia	0.062270	0.0235	2.653	0.00825	**
Region Middle East & NAfr EEastEsast&NENA	0.064940	0.0224	2.901	0.0039	**
Region The Amers	0.054300	0.0213	2.545	0.01127	*
Region Asia & Pacific	0.040820	0.0206	1.981	0.04823	*
R2 = 0.223408, N = 508

Number of obs: 508, groups: Year, 6; Significance: *: p < 0.05, **: p < 0.01, ***: p < 0.001.

.

Panel data analysis and linear mixed models are widely used in social science, economics, business analytics, and in many other fields of science. The fixed-effect and random-effect estimations of these models reveal important features of the relationship between the dependent and the independent variables. Fixed effects account for variables or factors that remain constant across observations, in our case across countries.

Random effects are used to account for variability and differences between different entities, which can be countries or time periods. In our model, the random effects were estimated for the various years.

The model estimates show the influence of a unit changes in the independent variables on the NRCA value, on average, and the random intercept values modify this estimated value by adding the relevant constant for each year. This means, using the results in Table 5, that the following equation can estimate NRCA for the Asia and Pacific region in 2017 (keeping only the significant estimates in the equation):

NRCA = −0.001205 WHSC_no + 0.000003 KnownSpecies + 0.005909 WHSN_no + 0.009548TServInf − 0.003565 EnvTreaty + 0.040820 − 0.0010744230

This type of modelling is particularly suitable in panel analyses, when we work with a large group of countries (or individuals, subjects) and have multiple-year observations for them. The fixed-effects estimations in these models can estimate the impacts of time-invariant factors, and the random effect estimations account for variations in outcomes that cannot be explained by observed variables alone. Thus, the fixed effect of a variable can be considered as the average estimate of its impact, while the random effect adds the individual variations to this. This type of modelling technique is very popular in cross-country economic comparisons of several years, and is also favoured in sociology and psychology when repeated observations are taken about a large group of individuals.

Model diagnostics require the testing of the residuals for normality, and it was approved by visual analysis as well as by the Kolmogorov–Smirnov test (see the first panel in Appendix B). As the p-value of the statistical test is higher than 0.05 (p = 0.0768), the normality of the residual series cannot be rejected.

The model suggests that NRCA is negatively influenced by WHSC and EnvTreaty values and positively influenced by TServIng, KnownSpecies, WHSN, and each of the various regions, though to different extents. The random effects of the years are also visible, though their magnitude is less than 10% of the regional fixed effects. The positive impacts of the factors are reasonable, but the negative effects are difficult to explain, as cultural heritage should reasonably be considered as a tourist attraction, and environmental treaties should also enhance a country’s image as devoted to the protection of natural heritage. The correspondence between observed and predicted NRCA values is not too strong (R = 0.474771 and R² = 0.223408), and this further supports the need for more refined analysis. As the regions show significant impacts, this suggests that the analysis could be performed by region.

To analyse regions separately, the Region variable was removed from the model structure, and the fixed effects of the years were added to see if there were considerable differences between years. Each region has several—but varying—significant fixed slope effects, but the random intercepts are at the 10⁻¹⁹ to 10⁻¹⁶ range, so their impact is negligible.

Table 6. Regionwide analysis by panel regression.

Fixed Effects:	Sub-Saharan Africa					Random Intercept
N = 75, R2 = 0.61468	Estimate	Std. Error	t Value	Pr(>|t|)		(Year)
WHSC_no	0.000672	0.000316	2.125000	0.037660	*	2011: 8.590075 × 10−19
ProtAreaPct	−0.000034	0.000045	−0.749000	0.456930		2013: 1.040383 × 10−17
KnownSpecies	0.000000	0.000000	0.973000	0.334540		2015: 3.467942 × 10−18
WHSN_no	0.000761	0.000477	1.595000	0.115950		2017: 8.538866 × 10−18
ForestPct	0.000010	0.000030	0.327000	0.745070		2019: −9.517754 × 10−18
RedListInd	−0.007073	0.002521	−2.805000	0.006730	**	2022: −3.155572 × 10−18
TServInf	0.001364	0.000508	2.685000	0.009320	**
EnvTreaty	−0.000138	0.000125	−1.102000	0.274850
factor(Year)2011	0.022120	0.004432	4.990000	0.000005	***
factor(Year)2013	0.003733	0.003357	1.112000	0.270550
factor(Year)2015	0.004210	0.003350	1.257000	0.213700
factor(Year)2017	0.003442	0.003174	1.085000	0.282370
factor(Year)2019	0.003335	0.003197	1.043000	0.300990
factor(Year)2022	0.002173	0.002870	0.757000	0.451830
Fixed effects:	Europe and Eurasia					Random intercept
N = 198 R2 = 0.36054	Estimate	Std. Error	t value	Pr(>|t|)		(Year)
WHSC_no	−0.001283	0.000401	−3.196	0.001638	**	2011: 2.783644 × 10−17
ProtAreaPct	−0.000247	0.000317	−0.78	0.436233		2013: −4.839659 × 10−17
KnownSpecies	0.000031	0.000007	4.326	2.49 × 10−5	***	2015: 1.156400 × 10−16
WHSN_no	−0.000544	0.002359	−0.231	0.817764		2017: −1.007347 × 10−18
ForestPct	0.000500	0.000223	2.241	0.026229	*	2019: −3.525714 × 10−17
RedListInd	−0.277600	0.071520	−3.882	0.000144	***	2022: −3.449263 × 10−17
TServInf	0.006441	0.003723	1.73	0.085327	+
EnvTreaty	−0.004024	0.001256	−3.204	0.001596	**
factor(Year)2011	0.297100	0.071990	4.127	0.999989
factor(Year)2013	0.291200	0.071390	4.079	0.999989
factor(Year)2015	0.303000	0.071080	4.263	0.99999
factor(Year)2017	0.262500	0.074210	3.537	0.999988
factor(Year)2019	0.263500	0.073840	3.569	0.999988
factor(Year)2022	0.292200	0.073770	3.962	0.999988
Fixed effects:	Middle East and North Africa					Random intercept
N = 47, R2 = 56136	Estimate	Std. Error	t value	Pr(>|t|)		(Year)
WHSC_no	−0.00169	0.00119	−1.41700	0.16580		2011: 4.250664 × 10−19
ProtAreaPct	−0.00058	0.00033	−1.76800	0.08630	+	2013: −8.032700 × 10−19
KnownSpecies	0.00001	0.00001	1.61100	0.11660		2015: 9.908962 × 10−19
WHSN_no	0.01544	0.00984	1.56800	0.12630		2017: 1.213061 × 10−18
ForestPct	0.00187	0.00083	2.24200	0.03180	*	2019: 8.087071 × 10−19
RedListInd	−0.08722	0.05476	−1.59300	0.12070		2022: 1.455326 × 10−18
TServInf	0.00274	0.00349	0.78600	0.43760
EnvTreaty	0.00390	0.00172	2.26500	0.03020	*
factor(Year)2011	−0.01478	0.03980	−0.37100	0.99950
factor(Year)2013	−0.01176	0.03729	−0.31500	0.99960
factor(Year)2015	−0.00489	0.03742	−0.13100	0.99970
factor(Year)2017	−0.01893	0.03707	−0.51100	0.99960
factor(Year)2019	−0.01691	0.03714	−0.45500	0.99960
factor(Year)2022	−0.01592	0.03737	−0.42600	0.99960
Fixed effects:	The Americas					Random intercept
N = 96, R2 = 0.74999	Estimate	Std. Error	t value	Pr(>|t|)		(Year)
WHSC_no	−0.00390	0.00073	−5.375	7.04 × 10−7	***	2011: −9.632035 × 10−18
ProtAreaPct	0.00059	0.00042	1.41	0.16245		2013: −6.476665 × 10−18
KnownSpecies	0.00000	0.00000	0.164	0.87038		2015: −1.050447 × 10−17
WHSN_no	0.01059	0.00185	5.731	1.60 × 10−7	***	2017: −8.136377 × 10−18
ForestPct	−0.00049	0.00027	−1.821	0.07228	+	2019: −9.595060 × 10−18
RedListInd	−0.14170	0.04569	−3.101	0.00264	**	2022: −5.729711 × 10−18
TServInf	0.00309	0.00448	0.69100	0.49187
EnvTreaty	−0.00944	0.00143	−6.605	3.70 × 10−9	***
factor(Year)2011	0.26150	0.05065	5.163	0.99892
factor(Year)2013	0.25250	0.05056	4.994	0.99893
factor(Year)2015	0.27020	0.05099	5.298	0.99889
factor(Year)2017	0.28570	0.05286	5.405	0.99872
factor(Year)2019	0.28340	0.05304	5.344	0.99871
factor(Year)2022	0.29800	0.05159	5.776	0.99882
Fixed effects:	Asia and Pacific					Random intercept
N = 92, R2 = 0.43755	Estimate	Std. Error	t value	Pr(>|t|)		(Year)
WHSC_no	−0.00213	0.00096	−2.21300	0.02986	*	2011: 2.915842 × 10−19
ProtAreaPct	0.00187	0.00067	2.76600	0.00708	**	2013: −6.460514 × 10−19
KnownSpecies	0.00001	0.00000	2.60000	0.01114	*	2015: −1.761062 × 10−19
WHSN_no	0.00189	0.00241	0.78300	0.43619		2017: −4.110190 × 10−19
ForestPct	−0.00124	0.00030	−4.11700	0.00009	***	2019: −1.370761 × 10−19
RedListInd	−0.12640	0.06485	−1.94900	0.05491	+	2022: −3.720638 × 10−19
TServInf	0.00309	0.00448	0.69100	0.49187
EnvTreaty	−0.00169	0.00274	−0.61600	0.53948
factor(Year)2011	0.15490	0.08204	1.88800	0.95579
factor(Year)2013	0.13190	0.08173	1.61400	0.95824
factor(Year)2015	0.13110	0.08326	1.57500	0.95592
factor(Year)2017	0.11630	0.08667	1.34200	0.95212
factor(Year)2019	0.12580	0.08719	1.44200	0.95006
factor(Year)2022	0.09867	0.08714	1.13200	0.95392

Significance: +: p < 0.1, *: p < 0.05, **: p < 0.01, ***: p < 0.001.

presents the results by region.

The diagnostic testing of the residuals justifies their reasonable normality, though occasionally this required the deletion of a few outliers (see Appendix B). As the diagnostics and the R² values indicate, all regional models provide reliable estimates.

Table 7. Summary of significant estimates of fixed effects by regions.

	Europe and Eurasia	The Americas	Asia and Pacific	Sub-Saharan Africa	Middle East and North Africa
TServInf	0.00644	0.01357		0.00136
WHSC_no	−0.00128	−0.00390	−0.00213	0.00067
ProtAreaPct			0.00187		−0.00058
KnownSpecies	0.00003		0.00001
WHSN_no		0.01059
ForestPct	0.00050	−0.00049	−0.00124		0.00187
RedListInd	−0.27760	−0.14170	−0.12640	−0.00707
EnvTreaty	−0.00402	−0.00944			0.00390
factor(Year)2011				0.02212

sums up the main findings, presenting only the significant fixed effects of the heritage components. It is quite clear that the regions considerably differ in the following aspects:

• The Year factor did not show any significant effects except for the first year (2011) in Sub-Saharan Africa.
• The tourism service infrastructure (TServInf) had a positive effect in Europe and Eurasia, the Americas, and Sub-Saharan Africa, but no impact was found in the Middle East and North Africa or in the Asia and Pacific region.
• The number of cultural world heritage sites (WHSC_no) had a negative significant effect except in Sub-Saharan Africa, where a positive impact was noted, and in the Middle East and North Africa, where no effect was found.
• Regarding the natural heritage components:

  ○

  The Red List Index had a significant negative impact everywhere except the Middle East and North Africa.

  ○

  The forest area proportion (ForestPct) was positive in Europe and Eurasia and in the Middle East and North Africa and negative in the Americas and in the Asia and Pacific region.

  ○

  The number of ratified environmental treaties (Envreaty) showed significantly positive impacts in the Middle East and North Africa, and negative impacts in Europe and Eurasia and in the Americas.

  ○

  The proportion of protected areas (ProtAreaPct) was significantly positive in the Asia and Pacific region and significantly negative in the Middle East and North Africa region.

  ○

  The number of known species had significant positive impacts in Europe and Eurasia and in the Asia and Pacific region.

  ○

  The number of natural world heritage sites (WHSN_no) had a positive impact only in the Americas.

As various heritage components have different magnitudes, the coefficients were compared, multiplying them by the relevant regional mean values; thus, their values are made comparable (Figure 2, top panel). The bottom panel of Figure 2 presents these impacts in percentages, with the full impact scaled to 100 percent in each region.

The impact of the Red List Index seems to be the largest, representing approximately 58% of the impacts in Europe and Eurasia, 52% in Sub-Saharan Africa, 47% in Asia and Pacific, and 26% in the Americas, and these impacts are all negative. The single largest positive impact is seen in Middle East and North Africa, where the number of environmental treaties take up 86% of all impacts (with a positive sign), while this factor has a considerable importance in the Americas (−46%) and in Europe (−21%). The tourism infrastructure is rather influential in Sub-Saharan Africa (+ 33%), while its influence is much less in the Americas (+12%) and in Europe and Eurasia (+7%). The forest areas have a 23% influence, and the protected areas a 13% influence in the Asia and Pacific region, while the number of cultural world heritage sites show a 15% positive influence in Sub-Saharan Africa. The rest of the factors—natural world heritage sites, known species, and protected areas—play a much smaller role, with only 0% to 13% importance.

In this respect, the Americas and Europe and Eurasia seem to have much in common, while the Asia and Pacific region slightly differs from them. The Middle East and North Africa region and Sub-Saharan Africa seem to have rather unique impact structures towards NRCA, which is not surprising considering their unique natural conditions, development levels, and cultural endowments.

4. Discussion

The revealed comparative advantage of the tourism industry has been analysed widely for many countries, using various forms of the RCA index [33,34,47,48,49], while the NRCA index has been applied less frequently [34]. The present study applies NRCA, which is a new approach for a worldwide tourism analysis. The differences shown in Table 7 and Figure 2 are reflections of the strengths and weaknesses of various regions and the manifestations of these in international tourism export competitiveness [50].

The most influential factor on NRCA seems to be the Red List Index, and this has a considerable negative impact on all regions except the Middle East and North Africa, which is not very rich in species compared to the other regions. The negative impact of this on all regions means that the higher the index (i.e., the fewer species that are threatened by extinction), the less comparative advantage is shown by NRCA. This seems to suggest that tourists seek the sensational and are attracted to endangered sights; therefore, tourism exports increase where such unique living systems are found. This is true for all regions, except the Middle East and North Africa, where the mean number of known species is much lower than elsewhere; thus, wildlife is not a determinant factor in tourism. Increased tourist interest in regions with endangered wildlife was found in [51] about the wildlife of the Swedish Arctic and in [52] about ecotourism in general. The concept of wildlife tourism utilises this interest in observing and interacting with animals that may be endangered, threatened, or rare in worldwide destinations [53,54,55].

Other aspects of natural heritage also show diverse influences on NRCA. The positive impact of the number of known species (Europe and Eurasia, Asia and Pacific), the proportion of protected areas (Asia and Pacific), the proportion of forest areas (the Americas) and the number of ratified environmental treaties (Middle East and North Africa) are in line with our logical expectations, and their positive impact on tourism competitiveness is supported by previous research [31,56,57]. The more surprising findings in this respect are the negative impacts. The negative impacts of protected areas in the Middle East and North Africa may be due to their outstanding low level, which may make them less accessible to tourists. The negative effects of forest areas in the Americas and the Asia and Pacific region may be explained by their exceptionally high proportion, which may be a barrier to transport and travel. The negative impact of the number of ratified environmental treaties in Europe and Eurasia, the Americas, and the Asia and Pacific region may easily lead to increased costs related to environmentally threatened tourist attractions. Further research may be needed to clarify these impacts.

Tourism infrastructure is expected to have a significant impact on tourism market success, as better services increase tourism revenues [33,56,57], and, in certain destinations, tourism services infrastructure was found to be a more influential factor than natural or cultural endowments [48]. In our findings, the surprising fact is that tourism services infrastructure proved to be non-influential in the Asia and Pacific region and in the Middle East and North Africa. According to the last two WEF reports [15,16], these two regions performed the worst regarding the tourism services and infrastructure indicators, both in 2019 and 2022, and this may be the explanation why this factor could not contribute to the export competitiveness of the industry.

Cultural world heritage sites (WHSCs) are also influential factors in four of the five regions, but their impact is more controversial. They contribute to the increase of tourism export competitiveness only in Sub-Saharan Africa. Their impact is negative in Europe and Eurasia, in the Americas, and in the Asia and Pacific region, while in the Middle East and North Africa region, their impact is not significant. This can be explained by the fact that these sites are rather rare in Sub-Saharan Africa, so their existence is a comparative advantage for the countries possessing them, while, in the rest of the world, their prevalence is far higher. In the regions with negative impact, their presence may lead to overtourism features [50,58,59], leading to congestion, and may even lead to the deterioration of these sites or enhanced conservation costs that decrease their net positive impacts. Prior research indicated negative impacts of cultural heritage sites on RCA index values in Eastern Africa [33], but, when suitable infrastructure was available, this impact turned positive. Positive impacts of cultural world heritage sites were demonstrated on tourism arrivals and receipts in Europe and America [29,31], but these cannot be equated to tourism export competitiveness, and this points to the need to assess how tourism export competitiveness relates to the export competitiveness of other industries.

5. Conclusions

Based on the above results, the following conclusions may be drawn. First, the application of the NRCA index can reveal important features of the export competitiveness of the tourism industry and its regional differences. While RCA-based tourism analysis is frequent in the literature, the more reliable NRCA index is less applied for tourism analysis; therefore, the present research is an important contribution to the state of the art in this respect. Second, the analysis revealed that the destination competitiveness approach, based on the tourism endowments of destinations, usually shows a very different picture from the actual revealed comparative advantages, i.e., the market-based analysis of tourism performance. The panel regression pointed out that the main factors that are used in the evaluation of destination endowments do not always contribute positively to the export competitiveness revealed by the NRCA index. The regions considerably differ as to which endowments are successfully translated to market performance and which ones are more a drawback than a benefit. This points to the factors most valued by international tourists either for their abundance or for their actual uniqueness. There is a general tendency that what is most sought after tends to be overused, and, therefore, there is a danger that formerly appreciated resources may disappear and become no longer a positive factor in tourism performance.

Economic theory appreciates the economic importance of cultural heritage, as it provides positive externalities, enhancing employment, and improving human and social capital, while following the principles of sustainability. Cultural heritage provides a unique perspective on history and tradition and is an important component of cultural capital. As such, it is invaluable in educating younger generations, enhancing a sense of identity and belonging, critical thinking, and empathy towards diverse cultures. However, the phenomenon of overtourism poses a significant threat to heritage sites; therefore, risk-mitigation strategies should be developed to prevent damage to heritage. Such strategies could start with the careful analysis of the carrying capacity of heritage sites, followed by limiting rules on the access times, traffic restrictions favouring on-foot access, tax regulations, and higher prices of access and accommodation in the neighbourhood. The idea of tourism demarketing may serve the protection of such sites, though at the expense of lower economic benefit.

Based on the above, policy implications may be outlined, both for regions of developed tourism sectors and for those still underdeveloped in this respect. One such implication is that cultural heritage sites should be protected against overtourism in the most developed tourism regions, while they provide untapped opportunities in less developed areas. The same is true about environmentally threatened areas and the protection measures applied, which may become negative factors in tourism market success, especially when the natural (and cultural) resources are affected by overtourism. In spite of their negative impacts, these steps should be strengthened in the development of sustainable tourism, and new ways of maintaining and presenting these resources should be invented, utilising tourism innovations, such as artificial intelligence, virtual tours, and limited access to endangered resources.

The limitations of the present research include data availability, as only 117 countries could be covered for a 10-year-long period. More detailed analysis within and between region comparisons could also be performed, not stopping at the regional level but looking at sub-regions of continents or at climatic zones regardless of continental breakup. The same methodology can be applied to more localised areas, small, socio-economically similar groups of nations, to identify which direction tourism exports should take in the coming decades. The analysis could be performed on a continent-driven basis instead of a region-driven approach, to see the impact of regional differences within continents and regional similarities across continents. The influence of variables, such as a safety index, a digitalisation index, or access to online information, could also be used for grouping the countries, as they may also have an impact on tourism performance.

Tourism competitiveness can also be assessed by other indicators besides NRCA, and comparisons between indicators could also reveal interesting and unexpected patterns in export performance of the industry. These issues are areas of future research.