Destination competitiveness research over the past three decades: a computational literature review using topic modelling

ABSTRACT

The aim of this study is to comprehensively review the topics and themes studied in destination competitiveness research and to enrich tourism researchers’ literature review toolbox with the computational literature review technique. A novel computational literature review framework that can be applied to review tourism research domains characterised by an abundance of literature is developed. Enabled by the developed computational literature review framework with topic modelling, this study offers a comprehensive review of 1,130 destination competitiveness studies. The results identified several promising research topics, themes and theoretical perspectives that can be perused in future destination competitiveness studies.

KEYWORDS:

• Destination competitiveness
• computational literature review
• machine learning algorithm
• ordinary least squares regression
• topic modelling
• tourist destination
• tourism
• bertopic
• comprehensive literature analysis
• text mining

Introduction

Over the past three decades, researchers have published a large number of academic studies to explore the destination competitiveness phenomenon (Bahar & Kozak, 2007; S. Chen et al., 2023; Tsai et al., 2009; Xu et al., 2021). To consolidate the acquired knowledge and drive research progress, several literature reviews on destination competitiveness have been conducted (e.g. Y. R. Kim et al., 2022; Mior Shariffuddin et al., 2022). These existing reviews typically examined a limited subset of destination competitiveness studies, often less than 200 journal articles, from predefined research themes using the method of narrative literature review or manual systematic literature review. By focusing on pre-defined research themes, the past literature reviews were not able to provide a comprehensive overview of the research themes and topics studied in the broader destination competitiveness literature, which encompasses a wide range of sources, including journals, conference proceedings, books, and book chapters. However, such a comprehensive overview is crucial to understanding the development of destination competitiveness research over time in order to guide future endeavours (Karami et al., 2020).

A reason for the lack of comprehensive literature reviews covering the broad field of destination competitiveness literature might be the methodological challenge of extracting meaningful topics and synthesising thematic content from the large volume of diverse sources of destination competitiveness studies. Existing literature review methods, including manual systematic literature reviews and narrative literature reviews, are best suited for in-depth thematic analysis of a relatively small body of literature, because they rely entirely on the information processing capability of human researchers (Mortenson & Vidgen, 2016). To get a comprehensive overview of the themes and topics studied in the broad destination competitiveness literature and to identify promising directions that may advance the research landscape, novel literature review methods that can augment the information processing capability of human researchers are required. Existing destination competitiveness literature reviews often acknowledge these limitations (e.g. Cronjé & du Plessis, 2020), but have not yet addressed them.

This study aims to fill this research gap and enrich tourism researchers’ literature review toolbox by taking advantage of the recently developed computational literature review method that uses machine learning algorithms to identify thematic content from large volumes of academic studies (Antons et al., 2023). To comprehensively analyse the research themes and topics in the broad destination competitiveness literature using this novel literature review approach, we are guided by four key research questions (RQs). RQ1: “What are the important publication sources, publications and authors in destination competitiveness research?;” RQ2: “What research topics and themes have been studied in existing destination competitiveness literature;” RQ3: “What are currently prevailing research topics and themes in the destination competitiveness literature?,” and RQ4: “How have the popularity of research topics and themes changed in the destination competitiveness literature in the past three decades?.” By answering these research questions, our results complement existing literature with a comprehensive overview of the diverse sources of destination competitiveness research from several perspectives, such as popular research topics, themes, trends of these topics and themes, leading authors, and important studies.

The main contributions of this study are threefold. First, this study introduces a machine learning algorithm-based computational literature review framework that enriches tourism researchers’ toolbox for conducting literature reviews in research domains that are characterised by a large body of literature. Second, this study applies a topic modelling-based computational literature review framework to review the abstracts of 1,130 destination competitiveness studies, providing the most comprehensive descriptive analysis, topic and theme analysis of existing destination competitiveness literature so far. Third, this study identifies several promising directions that can be pursued in future destination competitiveness research.

Literature study

Destination competitiveness literature reviews

“Literature review is a summary of existing knowledge and an outlook on future research direction” (G. I. Huang et al., 2023, p. 1). It plays an integral role in advancing all research domains, including destination competitiveness, where several literature reviews have been conducted to drive research forward. For example, Abreu-Novais et al. (2016) reviewed the destination competitiveness literature on three research themes: definitions, theoretical models and measurement, providing insightful discussions that may inspire future research on each theme. Cronjé and du Plessis (2020) reviewed 121 publications from 1997 to 2018, offering key statistical information from the collected destination competitiveness literature, such as the continental distribution of case studies and research focus. Y. R. Kim et al. (2022) reviewed 119 publications from 2005 to 2021, revealing citation networks and research themes at different levels of competitiveness. Mior Shariffuddin et al. (2022) examined 80 publications from 1983 to 2001, detailing the distribution of key factors examined in previous destination competitiveness research. Xu and Au (2023) reviewed 183 publications from 2010 to 2022, providing an overview of the key research theories, research themes and factors from a demand and supply perspective.

Despite the contributions, existing literature reviews often examined a limited number of destination competitiveness studies from predefined perspectives/themes (see Table 1), while neglecting to provide a comprehensive overview of the broad destination competitiveness literature and details of the research themes and topics studied. However, such an overview is of great value in guiding future research endeavours (Kunc et al., 2018). The reasons for this research gap may lie in the literature review methods used, namely manual systematic literature review and narrative literature review, which are unable to extract thematic information from large volumes of academic publications. One potential literature review method in the tourism literature that can address this methodological challenge is bibliometric analysis, which is designed to examine the intellectual connections of research and the social connections of researchers in a large body of literature by analysing bibliographic data (Donthu et al., 2021).

Table 1. Summary of the analysed destination competitiveness literature reviews (N/A represents not available).

Literature	Method	No. Reviewed Publications	Time	Review Perspectives
Abreu-Novais et al. (2016)	Narrative Literature Review	N/A	N/A-2015	Definitions; Conceptual models; Measurements
Cronjé and du Plessis (2020)	Manual Systematic Literature Review	121	1997–2018	Literature Statistical Information
Y. R. Kim et al. (2022)	Manual Systematic Literature Review	119	2005–2021	Literature Statistical Information; Levels of Competitiveness
Mior Shariffuddin et al. (2022)	Manual Systematic Literature Review	80	1983–2021	Literature Statistical Information; Factors of Competitiveness
Xu and Au (2023)	Manual Systematic Literature Review	183	2010–2022	Literature Statistical Information; Research themes; Research Design (supply & demand perspectives); Theories; Destination Types

Although bibliometric analysis techniques, like co-words analysis, can explore research “topics” in the literature, the identified “topics,” such as “tourism” and “competitiveness” (Xu & Au, 2023) tend to be overly broad, due to the limitations in the underlying statistical mechanism (Mendiratta et al., 2023). To provide a comprehensive overview of the detailed research themes and topics explored in the broad destination competitiveness literature without being confined to a certain number and type of publication or research theme, this study therefore introduces computational literature review (Antons et al., 2023) – a novel machine learning algorithm-based literature review technique that has not yet been applied in tourism research. The following subsection details the method and advantages of computational literature review and compares it to traditional methods of literature analysis.

Computational literature review

The exponential growth of academic publications not only highlights the need for repeatable and systematic literature reviews but also makes reviewing difficult and labour-intensive (Mortenson & Vidgen, 2016). Given the inherent limitations of human information processing capacity, manual systematic literature reviews face challenges in processing and analysing large volumes of academic publications (Kunc et al., 2018). To address this challenge, the computational literature review method has recently been proposed. Computational literature review is structured processes that use text mining (Godnov & Redek, 2016) and natural language processing (Khurana et al., 2023) algorithms to comprehend large amounts of textual data to augment the information extraction capabilities of researchers to analyse the literature in a particular field (Antons et al., 2023). Unlike bibliographic analysis, which reviews large volumes of literature based on the bibliographic information of each study (Donthu et al., 2021), the computational literature review goes a step further by analysing the content of studies using machine learning algorithms (Antons et al., 2023). It focuses on extracting more detailed thematic content (Braun & Clarke, 2006) of studies as contained in their abstract or the full text, rather than on patterns reflected in the bibliographic data.

Compared to the traditional manual systematic literature review, computational literature review is more scalable and can be performed in near real-time. Its analysis is automated through machine learning algorithms and does not require direct human input (Kunc et al., 2018). Preliminary studies have demonstrated the effectiveness of this method in reviewing rich computer science and information systems literature (e.g. Karami et al., 2020; Mortenson & Vidgen, 2016), but it has not yet been implemented in tourism research. However, given the interdisciplinary nature of tourism research, where knowledge is shared not only in specific tourism journals but also in a range of publication sources from tourism-specific to discipline-specific outlets, the computational literature review is therefore of great value.

Antons et al. (2023) recommend six steps for conducting a computational literature review. The first step is to select a conceptual goal, which can be explicating, envisioning, relating, or debating, as defined by MacInnis (2011). Explicating involves explaining ideas and their relationships; envisioning involves identifying research gaps and conceiving new realities; relating involves identifying relationships between ideas and objects; debating involves challenging traditional beliefs about a research object. Once the conceptual goal is established, the second step is to collect and compile the corpus of literature for analysis. The third step is to select an appropriate AI algorithm according to the conceptual goals and available resources; examples of commonly used methods can be found in Antons et al. (2023). In the fourth step, the selected algorithm is applied to the corpus to perform computational analysis. In the fifth step, researchers use their domain knowledge to synthesise the algorithm outputs and transform them into meaningful knowledge. Finally, the generated knowledge is presented and visualised in a comprehensible form.

Given that the main purpose of this study is to examine the research themes and topics explored in the rich and diverse destination competitiveness literature to identify promising research directions for future studies (i.e. envisioning), a novel topic modelling-based computational literature review framework is therefore proposed. This framework follows the best practices of the existing computational literature reviews in computer science, information systems, and organisational studies (e.g. Antons et al., 2023; Karami et al., 2020; Mortenson & Vidgen, 2016).

Methodology

This section details the proposed computational literature review framework to achieve the conceptual goal of “envisioning” the destination competitiveness literature. The framework consists of three steps (see Figure 1): (1) literature collection, where the destination competitiveness literature is collected from academic databases; (2) statistical analysis, where the interesting statistical features of the collected literature are analysed; and (3) topic analysis, where the patterns of the detailed topics and themes explored in the destination competitiveness literature are analysed. The details of each step are presented in the following sections.

Figure 1. The proposed computational literature review framework for achieving the conceptual goal of envisioning.

Flowchart of the proposed computational literature review framework. It contains three parts, the left part represents the literature collection. It starts with two databases for literature collection and ends with cleaned articles. The top right represents the statistical analysis, it includes descriptive analysis and impact analysis. The bottom right represents the topic analysis, it includes distribution analysis and trend analysis.

Literature collection

The literature collection in the computational literature review follows the practice of traditional manually systematic literature reviews, starting with a keyword search in academic databases (Karami et al., 2020). In this study, Scopus ¹ (D1) and Web of Science ² (WOS) (D2) were used due to their large number of high-quality publications (Mortenson & Vidgen, 2016). Keywords commonly used in existing literature reviews: “tourism competi*,” “destination competi*,” and “destination comp* advantage,” were used to retrieve publications in the two databases (Y. R. Kim et al., 2022). In order to provide a comprehensive review of the literature, there was no restriction on the year of publication. The literature search was conducted in April 2023 and resulted in 3,887 publications (980 in Scopus and 2,907 in WOS) in journals, conference proceedings, books, and book chapters. After removing duplicates, 3,111 publications remained (D3). The abstracts, publication sources, authors, titles, and citations of these publications were extracted for subsequent analysis, following the approach of previous computational literature review studies (e.g. Karami et al., 2020).

In the eligibility evaluation process, 228 publications in predatory journals³ were removed to ensure that the literature review was based only on widely recognised and accepted knowledge. In the second step of the evaluation process, the abstracts of the remaining 2,883 publications were skimmed by the authors to remove those not relevant to tourist destination competitiveness. Two of the authors independently check whether a publication is related to destination competitiveness or not, disagreements are solved by discussion. This left 1,130 eligible publications (D4). Data cleaning was then performed on the 1,130 publications by removing the copyright information from the collected abstracts, transforming author names into a consistent format, and manually filling in occasional missing citations with those from Google Scholar.

Statistical analysis

Statistical analysis in our framework contains both descriptive analysis and impact analysis. The descriptive analysis provides an overview of the annual number of publications and their distribution across different publication sources in the field of destination competitiveness. This analysis provides researchers with an insight into the overall popularity of destination competitiveness research and helps them to identify important publication venues for future research. The impact analysis identifies the most influential publications and authors in the field of destination competitiveness. It serves to highlight scholars and publications that have made significant contributions to the field and provides a valuable guide for researchers wishing to familiarise themselves with the key destination competitiveness research and scholars.

In our framework, the impact of a publication is measured by the number of citations received, which is a widely accepted criterion in academia (Mortenson & Vidgen, 2016). The impact of a scholar is measured by the total number of citations and the h-index (Hirsch, 2005) in the collected literature. The total number of citations offers an overall measure of a scholar’s impact on destination competitiveness research, while the h-index provides an overview of a scholar’s ongoing impact by considering both the quantity and quality of the publications in the field of destination competitiveness. Formally, the h-index is defined as the highest number of h for which a scholar has published h publications that have each been cited h times.

Topic analysis

Following the best practices of existing computational literature reviews that pursue the conceptual goal of envisioning (e.g. Mortenson & Vidgen, 2016), topic modelling is selected as our main literature analysis technique. Topic modelling is a powerful text mining technique for identifying topics and thematic information in text documents.⁴ It has been successfully applied in many computational literature reviews (Karami et al., 2020). In this study, the state-of-the-art topic modelling technique ― BERTopic (Grootendorst, 2022) ― is used to perform the literature analysis. BERTopic assumes that each document discusses a topic, and the representation of this topic, namely, the relevant words that constitute the topic, can be generated through a three-step clustering process. First, the pre-trained language model is used to transform the collected documents into embeddings. Second, the dimension reduction technique is applied to reduce the dimension of the document embeddings. Finally, a class-based variant of the TF-IDF algorithm is applied to extract the representation for each topic based on the clustering results (Grootendorst, 2022).

The clustering-based design of BERTopic eliminates the need to specify the number of topics, which is somewhat difficult in traditional topic modelling techniques. BERTopic provides various algorithms at each step to create coherent topic representations. Pre-trained language models such as Sentence Transforms and Genism are available for document embeddings. Dimension reduction algorithms such as UMAP and TruncatedSVD are supported. For topic representation, algorithms such as C-TF-IDF and C-TF-IDF + Normalisation are supported.⁵ In our study, the default structure of BERTopic is used, namely, “Sentence Transforms,” “UMAP,” and “C-TF-IDF,” as it has demonstrated satisfactory performance in a wide range of applications (Kastrati et al., 2023).

After generating the topic representation of each publication, two of the authors independently coded the topics by examining the topic words and representation documents of each topic. Once all the topics were obtained, the two authors further coded the related topics into themes using similar coding procedures. Disagreements were discussed and resolved by the entire team of authors. The distribution analysis and trend exploration were then carried out to gain a comprehensive understanding of the topics and themes explored in the literature. In the distribution analysis, the distribution of research topics and themes is first analysed to understand the popularity of the topics and themes explored in the overall destination competitiveness literature. This is followed by an analysis of the distribution of topics and themes in the literature published in the last three years, in order to understand the popularity of the topics and themes examined in current destination competitiveness research.

In the trend exploration, the Ordinary Least Squares (OLS) regression (Karami et al., 2020) is performed to capture the longitudinal changes in the popularity of research topics. The calculation of the OLS regression is based on the percentage of documents related to a specific topic in each year. Formally, this percentage is expressed as $\mathit{P}\left(\mathit{T}|\mathit{D}\right)$, where $\mathit{T}$ is the number of documents on that topic and $\mathit{D}$ is the number of publications in that year. The slope of the OLS regression indicates whether a topic is becoming cold or hot, with a positive slope indicating a hot trend and a negative slope indicating a cold trend. The $\mathit{P}-\mathit{value}$ of the OLS regression indicates whether the trend is significant or not. The trend of the majority of topics belonging to a theme is taken as the theme trend. The patterns identified in this trend analysis are valuable for understanding the history and evolution of destination competitiveness and for guiding future research.

Results

Publication distributions and impact analysis

Figure 2 presents the distribution of the analysed 1,130 publications on destination competitiveness. Since 1994, there has been a steady increase in the number of publications on the competitiveness of tourist destinations. The annual number of publications shows a gradual increase from 1994 to 2007, peaking at 11 publications in that year. However, the publication trend took a notable turn after 2007, experiencing a significant increase and eventually reaching 97 publications in 2022, contributing to a cumulative total of more than 1,000 publications. This trend underlines the growing research interest in destination competitiveness over the last decades, especially after the key year of 2007.

Figure 2. Distribution of destination competitiveness publications over time.

Histogram of the number of publications on destination competitiveness from 1994 to 2023. The higher the bar, the higher the number of publications. There is a significant upward trend after 2007.

Figure 3 illustrates the distribution of these publications across different sources. The results show that journals are the dominant dissemination channel, accounting for the highest share of destination competitiveness studies at 66.6%. Conference proceedings are another important channel, accounting for 27.3% of the total, while books and book chapters account for 3.2% and 2.8% respectively.

Figure 3. Distribution of publication among different sources.

Pie chart of the number of publications in books, book chapters, conference proceedings and journal articles. The majority of the publications appear in journals and conference proceedings.

Figure 4 displays the distribution of the number of publications in the top 15 journals that published the most papers on destination competitiveness. Tourism Management and Tourism Economics have published the highest number of papers on destination competitiveness (35), followed by Current Issues in Tourism (28). The large number of publications in these reputable journals articulates that destination competitiveness is an important topic of research in tourism.

Figure 4. Distribution of the number of publications in the top 15 journals with the most publications on destination competitiveness.

Horizontal bar chart of the number of publications in the top 15 journals publishing destination competitiveness studies. The longer the horizontal bar, the higher the number of publications.

Table 2 displays the five most cited destination competitiveness studies and Table 3 the most cited authors (ranked by h-index and total citations). The most cited study is by Buhalis (2000), which extended the definition of destination to a perceptual concept and discussed the marketing aspect of destination competitiveness. The second most cited studies are the commonly accepted conceptual models of destination competitiveness developed by Crouch and Ritchie (1999), and Dwyer and Kim (2003). Other studies among the most cited in the collected literature are authored by Enright and Newton (2004), who focused on the empirical application of Crouch and Ritchie’s (1999) model, and Crouch (2011), who added relative importance to the competitiveness factors of the Crouch and Ritchie’s (1999) model. The most cited authors in our collected literature include Larry Dwyer, Dimitrios Buhalis, Geoffrey I. Crouch, Chulwon Kim and J.R. Brent Ritchie.

Table 2. Top-5 publications ranked by citations.

Title	Authors	Year	Source	Cites
Marketing the competitive destination of the future	Buhalis, D.	2000	Tourism Management	1503
Tourism, competitiveness, and societal prosperity	Crouch, G.I. and Ritchie, J.R.B.	1999	Journal of Business Research	938
Destination competitiveness: Determinants and indicators	Dwyer, L. and Kim, C.	2003	Current Issues in Tourism	922
Tourism destination competitiveness: A quantitative approach	Enright, M.J. and Newton, J.	2004	Tourism Management	573
Destination competitiveness: an analysis of determinant attributes	Crouch, G.I.	2011	Journal of Travel Research	411

Table 3. Top-10 researchers ranked by corpus specific H-Index and citations.

Rank	Author	H-Index	Rank	Author	Citations
1	Dwyer, L.	14	1	Dwyer, L.	1917
2	Kozak, M.	8	2	Buhalis, D.	1803
3	Mihalic, T.	7	3	Crouch, G.I.	1488
4	Kubickova, M.	7	4	Kim, C.	1103
5	Ramón-Rodríguez, A.B.	7	5	Ritchie, J.R.B.	1046
6	Croes, R.	6	6	Mihalic, T.	949
7	Cvelbar, L.K.	6	7	Enright, M.J.	853
8	Perles-Ribes, J.F.	6	8	Newton, J.	853
9	Moreno-Izquierdo, L.	6	9	Kozak, M.	535
10	Crouch, G.I.	5	10	Nijkamp, P.	426

Topics and themes distribution

Table 4 presents the 24 topics⁶ identified by BERTopic, together with examples of the representative documents and the resulting themes. Topic 1 groups together studies that identify key “attributes” that determine destination competitiveness, forming an independent theme. Topics 2 and 16 include studies that develop “evaluation models” of destination competitiveness. Specifically, topic 2 represents studies developing index systems to evaluate the competitiveness of destinations worldwide, while topic 16 represents studies developing composite indicators for destination competitiveness, exemplified by extensions to the World Economic Forum’s Travel and Tourism Competitiveness Index (TTCI). Topics 3, 4, 6, 9, 12, 14, 18, 20, 21, 22 and 23 represent studies that examine the impact of a “particular aspect” on destination competitiveness and the way in which the competitiveness of destinations can be improved by manipulating that aspect. The aspects explored in these topics are “information and communication technologies (ICT),” “hotels”, “sustainability,” “branding & marketing,” “transportation”, “tourist perceptions,” “network effect,” “cultural resources,” “pandemic & crisis,” “knowledge management” and “quality management” individually.

Table 4. Results of the detected topics and themes.

ID	Topic	Represent Documents	Resulting Theme
T1	Attributes	Azzopardi and Nash (2016); Crouch (2011)	Attributes
T2	Evaluation Index	S. Huang et al. (2017); Yang and Yu (2020)	Evaluation Method
T16	Composite indicators	Gómez-Vega and Picazo-Tadeo (2019); Mendola and Volo (2017)
T3	ICT	Boes et al. (2016); Milićević et al. (2020)	Particular Aspects
T4	Hotels	J. J. Kim et al. (2019); Xia et al (2019, 2020).
T6	Sustainability	Esparon et al. (2015); Pulido-Fernández et al. (2015)
T9	Branding & Marketing	Buhalis (2000); Pike and Mason (2011)
T12	Transportation	Khan et al. (2017); Łapko (2014)
T14	Tourist Perceptions	C. M. Chen et al. (2011); Jaković et al. (2017)
T18	Network Effect	Brandão et al. (2019); Romeiro and Costa (2010)
T20	Cultural Resources	Vašaničová and Košíková (2019)
T21	Pandemic & Crisis	Fernández et al. (2022); Xu et al. (2021)
T22	Knowledge Management	Amaya Molinar et al. (2021); García-Almeida and Klassen (2017)
T23	Quality Management	Go and Govers (2000); Rudančić-Lugarić (2014)
T5	Eco & Rural Destination	Chi and Han (2021); Thong et al. (2022);	Specific types of Destinations
T7	Health Tourism Destination	Mosammam et al. (2019); Yu et al. (2021)
T8	Croatia	Kulas et al. (2014); Miličević et al. (2017)	General Destinations
T10	Slovenia	Gomezelj (2011); Gomezelj and Mihalič (2008)
T11	Ukraine & Russia	Andrades and Dimanche (2017); Kovalov et al. (2017)
T13	Romania & Bulgaria	Nica et al. (2013); Popescu and Plesoianu (2019)
T15	South Africa & Botswana	du Plessis and Saayman (2018); Nare et al. (2017)
T17	Countries	Leung and Baloglu (2013); Stanišić et al. (2022)
T19	Latin America	Luisa Vásquez et al. (2021); Miller et al. (2008)
T24	DEA	Herrero-Prieto (2017); Martín et al. (2017)	Method

Topic 5 and topic 7 represent studies exploring the factors and evaluation methods of the competitiveness of destinations in specific types of tourism. The types of destinations examined in these two topics are eco-tourism and rural tourism destinations, and health tourism destinations (including medical and wellness tourism). Topics 8, 10, 11, 13, 15, 17 and 19 represent studies that examine the competitiveness of “general destinations,” which are in the form of countries or a specific region comprising several countries. Examples include Croatia, Slovenia, Ukraine & Russia, Romania & Bulgaria, South Africa, Botswana, European Union countries, Sub-Saharan Africa, and Latin America. Topic 24 represents studies that use data envelopment analysis (DEA) to evaluate destinations’ competitiveness, forming an independent theme – “method.” DEA is a popular non-parametric method that measures the efficiency of decision-making units. It represents studies that use an efficiency lens to understand destination competitiveness (Cracolici et al., 2008).

Table 5 shows the popularity of the respective topics and themes in the destination competitiveness literature, which is determined by the normalised frequency of each topic and the total normalised frequency of the topics belonging to each theme. The topic garnering the highest level of attention is “attributes” (T1) which significantly outweighs other topics. The most popular studies in this line of research are the conceptual model developed by Crouch and Ritchie (1999) and Dwyer and Kim (2003), which inspired many of the subsequent studies. Other popular topics include the development of the “evaluation index” (T2) and the role of “ICT” (T3) in influencing and contributing to destination competitiveness. The research theme that has attracted the most attention is the influence and interaction of “particular aspects” on destination competitiveness, followed by “attributes” of destination competitiveness, the competitiveness of “general destinations” in the unit of countries, and the “evaluation model” of destination competitiveness. In contrast, the themes of the competitiveness of “specific types of destinations” and the use of the DEA “method” are less discussed.

Table 5. The popularity of the detected topics and themes.

ID	Topic	Normalised Topic Frequency	Theme	Normalised Theme Frequency
T1	Attributes	0.249	Attributes	0.249
T2	Evaluation Index	0.104	Evaluation Method	0.126
T16	Composite indicators	0.022
T3	ICT	0.056	Particular Aspects	0.329
T4	Hotels	0.047
T6	Sustainability	0.042
T9	Branding & Marketing	0.034
T12	Transportation	0.032
T14	Tourist Perceptions	0.031
T18	Network Effect	0.019
T20	Cultural Resources	0.018
T21	Pandemic & Crisis	0.018
T22	Knowledge Management	0.017
T23	Quality Management	0.015
T5	Eco & Rural Destination	0.047	Specific types of Destinations	0.085
T7	Health Tourism Destination	0.038
T8	Croatia	0.036	General Destinations	0.197
T10	Slovenia	0.034
T11	Ukraine & Russia	0.033
T13	Romania & Bulgaria	0.032
T15	South Africa & Botswana	0.024
T17	Countries	0.019
T19	Latin America	0.019
T24	DEA	0.014	Method	0.014

Figure 5 visualises the distribution of topics and themes over the last three years (from 2021 to 2023), which represent the topics and themes that are currently being pursued by destination competitiveness researchers. The normalised frequency of each theme is denoted as a fuchsia dot. The dashed line is used to help understand the margin between the popularity of different themes. Currently, the most popular topic is still “attributes” (T1), followed by “pandemic & crisis” (T21), which has been brought to the forefront due to the impact of the COVID-19 pandemic on the tourism industry. The development of the competitiveness “evaluation index” (T2) and understanding destination competitiveness through “tourist perceptions” (T14) are also popular topics that are currently being pursued by destination competitiveness researchers. The research theme that currently attracting the most attention is “particular aspects” of destination competitiveness. This is followed by “attributes” and “general destinations,” while “specific types of destinations” and “method” are less discussed. This pattern is similar to that of the overall literature on destination competitiveness, indicating that the overall popularity of research themes in the destination competitiveness literature has not changed too much.

Figure 5. Distribution of the topics and themes from 2021 to 2023.

The normalised frequency of topics and themes from 2021 to 2023, including horizontal bars, fuchsia dots and dashed lines. The horizontal bars represent the normalised frequency of topics. The fuchsia dot represents the normalised frequency of themes. The longer the horizontal bar, the higher the normalised frequency. The dashed lines are used to visualise the margin between the normalised frequencies of the themes.

Topics and themes trends

Table 6 exhibits the trends of the detected topics and themes from 1994 to 2022. Year 2023 was excluded as publications were still ongoing in that year, the current available data can lead to spurious decreasing trends. The slope and p-value of the OLS regression are calculated using Python statsmodels.⁷ Of all the topics, 12 of them show a statistically significant increase trend (i.e. positive slope and p-value ≤0.05), including “Ukraine & Russia” (T11), “Croatia” (T8), “Slovenia” (T10), “South Africa & Botswana” (T15), “countries” (T17), “ICT” (T3), “tourist perceptions” (T14), “hotels” (T4), “pandemic & crisis” (T21), “transportation” (T12), “network effect” (T18), and “composite indicators” (T16). This indicates that these topics have gradually gained research attention and popularity over the past decades. Seven topics show an upward trend (not significant), including “Romania & Bulgaria” (T13), “eco & rural destination” (T5), “health tourism destination” (T7), “cultural resources” (T20), “knowledge management” (T22), “evaluation index” (T2) and “DEA” (T24). Only “attributes” (T1) shows a significant downward trend. This means that the topic may be well discussed and is gradually losing popularity. Four topics demonstrate a downward trend but are not statistically significant, including “Latin America” (T19), “quality management” (T23), “sustainability” (T6), and “branding & marketing” (T9).

Table 6. Trends of the research topics and themes from 1994 to 2022 (^ns p > 0.05; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001).

Topic (ID)	Slope (Topic trend)	P-value (Topic)	Theme	Trend
Ukraine & Russia (T11)	0.0014	***	General Destinations	⇈*
Croatia (T8)	0.0014	***
Slovenia (T10)	0.0012	***
South Africa & Botswana (T15)	0.0010	***
Countries (T17)	0.0008	***
Romania & Bulgaria (T13)	0.0009	ns
Latin America (T19)	−0.0062	ns
Eco & Rural Destination (T5)	0.0013	ns	Specific types of Destinations	⇈
Health Tourism Destination (T7)	0.0007	ns
ICT (T3)	0.0023	***	Particular Aspects	⇈*
Tourist Perceptions (T14)	0.0013	***
Hotels (T4)	0.0017	**
Pandemic & Crisis (T21)	0.0008	**
Transportation (T12)	0.0011	*
Network Effect (T18)	0.0006	*
Cultural Resources (T20)	0.0007	ns
Knowledge Management (T22)	0.0001	ns
Quality Management (T23)	−0.0044	ns
Sustainability (T6)	−0.0012	ns
Branding & Marketing (T9)	−0.0005	ns
Composite indicators (T16)	0.0009	**	Evaluation Method	⇈*
Evaluation Index (T2)	0.0027	ns
Attributes (T1)	−0.0122	*	Attributes	⇊*
DEA (T24)	0.0005	ns	Method	⇈

Looking at the evolution of destination competitiveness research at a more aggregated level by analysing the changes in themes over time reveals that “attributes” is the only theme that decreasing significantly, indicating the maturity of research in this theme. The theme “particular aspects” displays a significantly increased trend, with the majority of topics increasing significantly and some topics decreasing while not significantly. This indicates that understanding the competitiveness of destinations in terms of particular aspects is an important direction for research. Although some of the previously well-examined aspects are becoming less popular (e.g. quality management), new aspects have emerged (e.g. ICT, pandemic and crisis) and are driving this research direction. The themes “general destination” and “evaluation method” also show a significant upward trend. This suggests that the development of competitiveness evaluation methods and the study of the competitiveness of destinations worldwide are important research directions in the literature and have garnered considerable attention. The themes “specific types of destinations” and “method” demonstrate an increasing but not significant trend, which may be due to the fact that the topics in these two themes are just emerging and have not yet attracted enough attention.

Result discussion and implications

Implications for destination competitiveness literature

The large number of publications included in the analysis allows us to present the most comprehensive literature review on destination competitiveness thus far. To the best of our knowledge, this is the first literature review on destination competitiveness that includes thousands of publications from a variety of publication sources. Our results complement existing literature reviews by providing insights from various perspectives, especially concerning the detailed research topics and themes in the broad destination competitiveness literature – knowledge that cannot be captured by traditional tourism literature review methods. More specifically, the results of the descriptive analysis contribute to the literature by providing novel insights into the distribution of publications across different sources. Our results show that conference proceedings are also important sources of knowledge on destination competitiveness.

The impact analysis contributes to the existing literature with a list of highly cited publications and authors in the field of destination competitiveness, which can be understood as seminal works or highly contributing researchers, respectively. These results can help emerging researchers (e.g. PhD students) or researchers new to the field familiarise themselves with well-accepted knowledge and important researchers (Mortenson & Vidgen, 2016). Our results show that highly cited publications on destination competitiveness are conceptual models developed by Crouch and Ritchie (1999) and Dwyer and Kim (2003). This observation is consistent with previous literature reviews that identify these two models as key knowledge on destination competitiveness (e.g. Abreu-Novais et al., 2016; Y. R. Kim et al., 2022). Our results complement existing knowledge by showing that the most cited publication on destination competitiveness is Buhalis (2000), which defines the concept of destination and discusses the marketing perspective of destination competitiveness.

The results of the topic analysis first articulate 24 topics explored in the broad destination competitiveness literature and provide representative publications on each topic. These results enhance the existing literature by spotting detailed topics explored by destination competitiveness researchers, offering insights for auditing the research contributions made in this field. In addition to the broad topics, such as “tourism,” “tourists” and “competitiveness” identified in existing literature reviews through bibliometric analysis (e.g. Xu & Au, 2023), our study offers more in-depth information on topics that help better understand the detailed development of the research landscape. The results of the topic distributions further provide insights into the popularity of the topics and clarify the currently prevailing topics in the destination competitiveness literature. The related findings provide direct guidance to help researchers align their research topics, allowing them to focus on prevailing topics for incremental contributions or on less-explored topics for innovative contributions. For example, researchers may choose to explore hot topics, such as the influence of tourist perceptions on destination competitiveness (C. M. Chen et al., 2011; Tsai & Fong, 2021), or cold topics such as the influence of quality management on destination competitiveness.

The topic trends identified provide a longitudinal overview of the main topics discussed in previous destination competitiveness studies, which guides future studies by clearly indicating which topics have cooled off and which topics are becoming popular. Our results show that ICT is becoming an important focus of research on destination competitiveness, aligning with the results of Xu and Au (2023). Our results contribute to the existing literature with trends on other important topics that have not been revealed in the literature. For example, our results show a significant decrease in research examining the “attributes” that determine destination competitiveness or measures it from an “attributes” perspective, although the topic is still popular at the moment. This may be because the overuse of this typical perspective on destination competitiveness has raised researchers’ concerns about the novelty of this theoretical lens. Recently, Y. R. Kim et al. (2022) criticised that most of the existing destination competitiveness literature is merely a replication of the models developed by Crouch and Ritchie (1999) and Dwyer and Kim (2003) in different contexts. Therefore, future studies could try to introduce new theoretical angles to understand the destination competitiveness phenomenon, such as the customer experience (Carneiro et al., 2021; Sthapit et al., 2023; Tsaur et al., 2022; Xia et al., 2023) or the theory of change suggested by Xu and Au (2023).

The themes emerged and the trends of these themes provide a macro-overview of the existing destination competitiveness research. Most existing literature reviews on destination competitiveness are developed at this level, with themes predefined by researchers. Our study contributes to the existing literature with data-driven themes, which are inherently less prone to bias as the identification of these themes is not influenced by the researchers’ prior knowledge (Mortenson & Vidgen, 2016). Furthermore, our study contributes to the existing literature with trends of themes, and mapping between topics and themes in the broad destination competitiveness literature. These patterns are difficult to capture using traditional literature review methods but are helpful in spotting research gaps and directing future endeavours.

For example, the strong increasing trends of the themes “particular aspect,” “evaluation method,” and “general destinations” indicate that the understanding of the destination competitiveness from specific aspects (e.g. destination marketing), developing competitiveness evaluation methods, and exploring the competitiveness of destinations worldwide have attracted increasing research attention in the past and will continue to be important in the future. Therefore, future research attention is warranted. The increasing trend of the theme “specific types of destinations” entails that there is no set of factors or a universal model that could be used to understand the competitiveness of all types of destinations (Dwyer & Kim, 2003; Goffi & Cucculelli, 2019). The investigation of the competitiveness of destinations by considering their unique characteristics could be a popular theme for future research.

The increasing trend of the “method” theme reveals the popularity of using efficiency (i.e. DEA) as the research lens in the destination competitiveness literature (e.g. Cracolici et al., 2008; Mendieta-Peñalver et al., 2018). The efficiency lens allows for easy comparison of competitiveness across destinations by assuming homogeneity of the tourism services (Abad & Kongmanwatana, 2015). However, as the tourism industry enters the experience-based market, tourists are becoming more sophisticated in their tastes and seeking heterogeneous, personalised experiences rather than homogeneous tourism services (Shoval & Birenboim, 2019). Therefore, the efficiency lens may not be sufficient to understand destination competitiveness. On this basis, we argue that complementing or even replacing the efficiency with other perspectives, such as the tourist experience (Croes & Kubickova, 2013), may be better suited to fully capture the competitiveness of destinations in the experience-based tourism market. In this sense, it is expected that more topics (i.e. methods) will emerge in future studies on the theme of “method.”

The mapping between the theme “general destinations” and related topics reveals that most of the existing literature evaluated destination competitiveness at the country level (Y. R. Kim et al., 2022). However, destination is a perceptual concept that does not only refer to countries or administrative areas (Buhalis, 2000), which means that future studies could put more effort into investigating the competitiveness of destinations other than countries. Such as individual small destinations that are part of a city, or destinations that are difficult to define from an administrative perspective. This kind of research will significantly extend the theoretical boundaries of the existing destination competitiveness literature. We acknowledge that this will be challenging, due to the difficulty of data collection and the necessity of exploring new theoretical concepts. The mapping between topics in the theme “specific types of destinations,” indicates that only limited types of destinations have been frequently explored in this theme. Therefore, future research in this theme could focus on other types of destinations, such as integrated resorts (Ji & Yang, 2022) and island destinations (Croes, 2011; Walker et al., 2021).

Methodological implications

The machine learning algorithms-based computational literature review method presented in this study contributes to the broader tourism research methodology. With the increasing contribution of the tourism industry to global social and economic development, academic research on tourism has increased exponentially over the years. Computational literature review provides tourism researchers with new opportunities to synthesise knowledge from the content of the large bodies of valuable tourism literature. Our proposed computational literature review framework provides a prototype that can be directly used by other tourism researchers to develop toolkits for identifying research gaps in their fields of interest to direct future studies (i.e. pursuing the conceptual goal of envisioning). For researchers pursuing conceptual goals other than envisioning in computational literature reviews, such as relating, our framework also provides certain methodological foundations; they only need to replace the topic modelling technique adopted in this paper with the one that could support the conceptual goal they are pursuing.

Following our framework, a transparent and reproducible literature review could be easily conducted in any tourism research area characterised by an abundance of literature. Compared to traditional literature review methods, computational literature review is more inclusive, which provides several advantages that could be used to advance future tourism research. For example, computational literature review can reduce the bias of the small literature corpus or single literature source used in traditional literature reviews, as it can easily review a large volume of publications from multiple sources. The diverse sources of publications included in the computational literature reviews also increased the credibility of the results. The continued emergence of machine learning algorithms also makes the computational literature review more versatile than traditional literature review methods such as bibliometric analysis and meta-analyses.

Conclusions

Since the ever-increasing competition among tourist destinations, destination competitiveness has emerged as a prominent topic of research in tourism (Gomezelj & Mihalič, 2008; Zhou et al., 2015). Despite extensive discussion, the research on destination competitiveness remains incomplete, due to the complexity and multifaceted nature of the phenomenon (Mior Shariffuddin et al., 2022). In order to audit the research themes and topics examined in the broad destination competitiveness literature, and enrich the tourism researchers’ literature review toolbox, this study introduced a novel machine learning algorithm – topic modelling – based computational literature review framework. The framework was then applied to the abstracts of 1,130 destination competitiveness publications in journals, conference proceedings, books, and book chapters.

The results of the descriptive analysis complement the existing literature with the distribution of the publications across different sources. The results of the impact analysis identify the most cited publications and researchers in the field of destination competitiveness. The results of the topic analysis offer insights into the detailed research topics and themes explored in the broad destination competitiveness literature. Such knowledge is difficult to capture using traditional literature review methods, due to their limitations in extracting thematic information from a large body of literature (Antons et al., 2023). By analysing the identified topics, themes, trends of topics and themes, as well as mapping between topics and themes, this study identified several novel promising research directions that can be pursued in future destination competitiveness studies. Examples include investigating the competitiveness of destinations beyond the definition of administrative areas and using the experience lens and theory of change to enrich the existing destination competitiveness literature. The computational literature review framework developed in this study also opens new opportunities for tourism researchers in fields other than destination competitiveness to consolidate the wealth of knowledge in their fields of interest.

Despite the insightful analysis and suggestions that may help progress the field forward, our study also has limitations. First, our literature review only covered English publications in academic databases. However, publications in other languages are also important for destination competitiveness research. A natural extension of this study would be to include destination competitiveness publications published in languages other than English in the analysis. Second, our computational literature review focused exclusively on the conceptual goal of envisioning on destination competitiveness literature. Future studies could complement this exploration by investigating other conceptual goals such as relating and debating (MacInnis, 2011). Finally, as mentioned by Antons et al. (2023), computational literature review is not a replacement but rather a complement of traditional literature review methods. These different literature review techniques can be used collectively to advance the understanding of destination competitiveness (or other research domains of interest). Therefore, another extension of this study could involve conducting mixed-method literature reviews by integrating our proposed computational literature review framework with other literature analysis techniques such as co-authorship analysis in bibliographic analysis (Donthu et al., 2021).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10548408.2024.2332278

Additional information

Funding

This research is supported by an Australian Government Research Training Program (RTP) Scholarship; and the collaboration is partially supported by one SRG grant from the University of Macau.

Notes

1. https://www.scopus.com/

2. https://webofscience.com/

3. https://predatoryreports.org/

4. The documents in this study refer to the abstracts of the collected publications.

5. Details of the supported algorithms in BERTopic are available at https://maartengr.github.io/BERTopic/.

6. The word representations of each topic are provided in Table A1 of the Appendix.

7. https://www.statsmodels.org/stable/generated/statsmodels.regression.linear_model.OLS.html