Decoupling China’s Tourism Economy from Carbon Emissions Through Digitalization: A Supply-Side Analytical Framework

Abstract

Global tourism activities have become increasingly digitalized, yet the economic and environmental impacts of digitalization on tourism remain underexplored. This study develops a supply-side analytical framework to examine whether and how digitalization decouples tourism economy from carbon emissions by integrating Ghosh input-output analysis, subsystem analysis, and structural decomposition analysis. Our findings reveal that digitalization has largely decoupled China’s tourism economy from carbon emissions, with the increases in economic gains notably outpacing those in emission losses. Specifically, the digital-enabled tourism value-added (DTV) increased by approximately 18 times from 2002 to 2017, while digital-enabled tourism emissions (DTE) only grew by about 11 times. Between 2017 and 2020, due to the impact of the COVID-19 pandemic, the DTV decreased by about 61%, and DTE dropped by approximately 63.5%. The expansion in DTV can be primarily attributed to advancements in software and IT services and electronic components, while the increase in DTE is significantly driven by software and IT services and communication services. The growth in digital supply emerged as the predominant driver for the surging DTV and DTE, with the emission intensity of tourism subsectors acting as a notable constraint. This study offers both a methodological framework and empirical evidence aimed at guiding policy initiatives that target the digitalization and low-carbon transition of the tourism sector.

1. Introduction

In recent years, the tourism sector—a pivotal contributor to the global economy—has witnessed significant expansion before the advent of the COVID-19 pandemic, with tourism receipts increasing by 3~5% annually, outpacing the global economic growth rate [1]. By 2019, tourism’s contribution to global GDP reached an unprecedented 10.4%, amounting to over ten trillion dollars [2]. Despite these economic achievements, the industry faces considerable environmental challenges, notably its substantial carbon footprint, which represents about 8~11% of worldwide emissions [1]. A substantial increase in carbon emissions from the sector is anticipated with the rebound of tourism activities in the post-pandemic era. This situation highlights the urgent need for strategies that promote the decoupling of tourism economic growth from carbon emissions [3].

The tourism industry has undergone a profound digitalization transformation, leveraging digital technologies such as online platforms, advanced analytics, intelligent applications, and virtual reality technology to revolutionize established norms and catalyze growth. This shift has significantly altered distribution strategies [4], enhanced operational efficiencies [5], streamlined costs [6], and elevated service quality alongside customer experiences [7] thus fueling economic advancement within the sector. Academic inquiries into the repercussions of digitalization on tourism have been comprehensive, accentuating its transformative potential on business operations and practice. Aghaei et al. (2021) explored the transformative potential of blockchain in redefining business models and distribution mechanisms [8]. Buhalis et al. (2022) scrutinized the embryonic integration of the metaverse in tourism, underscoring its profound potential to enrich tourist experiences and enable collaborative value creation [9]. Tang (2024) investigated the relationship between the digital economy and the productivity of Chinese tourism firms, finding that the digital economy enhances the total factor productivity of these enterprises by improving the market environment [10]. Wang et al. (2024) employed econometric methods to examine the impact of digital economy on China’s tourism carbon emissions [11].

From a macro-economic perspective, the impacts of digitalization on tourism have been mainly evaluated through econometric models. Adeola et al. (2019) employed a dynamic panel gravity model to dissect the nexus between digitalization and tourism development in Africa, discerning a positive association [12]. Likewise, Zhu et al. (2021) investigated how digital adoption influences the tourism engagement of rural Chinese populations, revealing a significant uptick in participation rates [13]. Wu et al. (2024) analyzed the impact of the digital economy on China’s tourism industry from 2011 to 2018, finding that the digital economy significantly accelerates tourism recovery [14].

While the economic implications of digitalization within the tourism sector are well-documented, its environmental impacts particularly regarding carbon emissions remain underexplored and warrant further investigation. Studies like those by Ma et al. (2021) posit that digitalization could potentially enhance tourists’ environmental awareness and streamline management processes, which may contribute to a reduction in carbon emissions [15]. Balsalobre-Lorente et al. (2023), using an Augmented Mean Group and two-step GMM method, concluded that the moderate effect of digital technology on tourism leads to a significant decrease in CO₂ emissions [16]. Wei et al. (2023) argued that digital technology can ultimately reduce resource consumption and thus decrease carbon emissions, contributing to sustainable tourism [17]. Conversely, Gössling (2021) concluded that information and communication technology can, at best, only marginally promote sustainability in tourism [18]. Nonetheless, the debate persists, with some scholars skeptical of digitalization’s environmental impact, noting the absence of robust empirical backing. Zhou et al. (2022) provide evidence of digitalization’s significant ‘push effects’ on emissions within the transportation subsector of tourism, a critical component of the industry’s infrastructure [19]. This ongoing debate underscores an ongoing debate regarding the efficacy of digitalization in emission mitigation within the tourism sector. Moreover, the precise extent and mechanisms by which digitalization affects carbon emissions within the tourism sector remain insufficiently understood. Therefore, it is imperative to conduct a comprehensive assessment of the simultaneous economic and environmental impacts of digitalization on tourism. Such an analysis is crucial for formulating strategic digital initiatives aimed at decoupling economic growth from carbon emissions, thereby guiding the sector toward sustainable development.

China’s tourism sector has seen substantial growth, with revenues surging from CNY452 billion in 2000 to CNY 6630 billion in 2019, marking an annual growth rate of 11.4% in 2019—a figure that significantly surpasses the global average and positions China at the forefront of the global tourism market [20,21]. Concurrently, carbon emissions attributed to China’s tourism industry have seen a nearly 10% annual increase from 1990 to 2019 [22], underscoring the pressing imperative for environmental sustainability measures. These measures are instrumental in realizing China’s ‘dual carbon’ goals of achieving peak carbon emissions and striving for carbon neutrality. In addition, China’s considerable share of the global digital output—over 30% as of 2020 [23]—accentuates the importance of examining the rapid digitalization within its tourism sector. With the tourism sector poised for a swift post-pandemic recovery, it is essential to assess the impacts of digitalization on both the value-added and carbon emissions of tourism. Investigating the potential of digitalization to decouple economic growth from carbon emissions in China’s tourism industry has become increasingly critical.

To address these identified gaps, this study quantitatively assesses the dual impacts of digitalization on the economic and environmental facets of the tourism sector. We develop a multi-phase supply-side analytical framework based on environmentally extended input-output analysis (IOA) (All the abbreviations appeared in this study are listed and explained in Appendix A.)—a technique proven effective in evaluating economic benefits and carbon emissions across different sectors and their interrelationships [24,25]. This framework facilitates an examination of the contributions of digitalization to value-added and carbon emissions within tourism. With the Ghosh IOA [26], we first quantify the influence of digital inputs on the tourism sector’s value-added and carbon emissions. Subsequently, subsystem analysis (SA) further enables the tracing of value-added and emissions flows from digital subsectors to their tourism counterparts. Finally, structural decomposition analysis (SDA) dissects how attributes of digital supply—i.e., scale, structure, and product type—influence changes in the economic and environmental outputs of tourism. We validate our methodological framework through a case study of China’s tourism and digital sectors, reflecting the nation’s significant strides in both domains in recent years.

This study aims to contribute to the literature on digitalization and sustainable tourism through a supply-side lens. First, we propose a novel, phased analytical framework that integrates IOA, SA, and SDA techniques. This framework assesses the economic and environmental impacts of digitalization within the tourism sector, tracing the sources, destinations, and drivers of these effects. Second, we examine the dynamics of digital subsectors and their influence on the economic and environmental performance of tourism subsectors. By identifying which digital subsectors promote economic growth and which predominantly drive emissions, we provide key insights for strategically leveraging digitalization to optimize specific aspects of the tourism industry. Third, we conduct a case study of China’s tourism and digital sectors, offering a reference model for evaluating the impact of digitalization in similarly digitized sectors and geographic contexts. Our findings inform policy formulation and strategic planning aimed at achieving the dual objectives of tourism development and carbon emission control in an era of increasing digital application.

2. Methodology and Data

This study constructs an integrated supply-side analytical framework to evaluate the economic and environmental effects of digitalization within the tourism sector. The methodology progresses through a sequence of five methodical stages: (1) identifying the key components of both the digital and tourism sectors; (2) analyzing the economic and carbon impacts enabled by digitalization on tourism; (3) quantifying the decoupling status of digital-enabled impacts; (4) exploring the flow of economic benefits and carbon emissions from digital subsectors to tourism subsectors; and (5) analyzing the underlying drivers of these impacts (Figure 1).

The first step of this framework is to define the core of the digital sector and the core of the tourism sector. The digital sector is broadly recognized to encompass both digital devices and services. By aligning with the sector classification outlined in national input-output (I-O) tables, the essence of the digital sector can be characterized by specific subsectors such as communication equipment, computers, electronic components, audio-visual apparatus, and other communication and electronic equipment, along with communication services, software, and IT services. Regarding the tourism sector, a universal consensus on its core and boundaries has yet to be established. According to the World Tourism Organization, the tourism industry is defined by principal products categorized under the Central Product Classification system. Common categories include travel agency services, accommodation, food and beverage services, transportation, sightseeing, entertainment, and shopping, among others, allowing for adjustments based on each country’s specific context. In line with China’s National Statistical Classification of Tourism and Related Industries [27] and the national I-O tables, the core of the tourism sector is identified as the subsectors directly offering products or services to tourists, such as catering, accommodation, transportation, sightseeing, shopping, entertainment, and comprehensive services. These compositions of the digital sector and tourism sector are respectively detailed in

Table A1. The core of the digital sector.

Sector Subdivision	Sector Clusters	Abbreviation
Digital devices	Communication equipment	Comm. Devices
	Computer	Computer
	Electronic components	E-Components
	Audio-visual apparatus	AV Apparatus
	Other communication and electronic equipment	Other Devices
Digital services	Communication services	Comm. Services
	Software and IT services	IT Services

and

Table A2. The core of the tourism sector.

Sector Subdivision	Sector Clusters	Abbreviation
Tourism catering	Catering services	Catering
Tourism accommodation	Hotels	Accommodation
Tourism transportation	Railway transportation; Road transportation; Water transportation; Air transportation; Urban public traffic	Transportation
Tourism sightseeing	Ecological protection; Management of public facilities	Sightseeing
Tourism shopping	Retail trade	Shopping
Tourism entertainment	Entertainment Cultural and art activities Sports activities	Entertainment
Comprehensive services	Postal services Business Services Tourist agency industry Financial service Leasing service Resident service and other services	Compr. Services

, Appendix B.

The second phase is to quantify the economic and carbon impacts enabled by digitalization through the application of the I-O method from a supply-driven perspective. Specifically, the Ghosh supply I-O model is utilized to measure digital-enabled tourism emissions (DTE) to digital-enabled tourism value-added (DTV). Here, DTE represents the carbon emissions generated in the tourism sector facilitated by the application of digital products, while DTV denotes the value-added originating from digital primary inputs. The Ghosh model accurately assesses the contributions of digitalization to tourism by examining the forward linkage effect, which encompasses the procurement of digital products for tourism services.

The third phase involves calculating the digital-enabled emission intensity (DEI), defined as the ratio of DTE to DTV. The DEI index, combining the Tapio decoupling index [28], can reveal the decoupling effect of digitalization on the added value and carbon emissions of the tourism sector. To ascertain the DEI for the total tourism sector and individual tourism subsectors, we calculate the DTE per unit of DTV within both the overall sector and within each subsector.

The fourth phase entails analyzing the flow of economic benefits and carbon emissions from digital subsectors to tourism subsectors. We employ SA to identify which digital subsectors are the primary sources of emissions and which offer the most significant economic benefits. This step also involves pinpointing the tourism subsectors that receive the highest emissions and those that gain the most substantial economic advantages from digitalization.

The fifth phase involves a deeper examination of the factors influencing changes in DTV, DTE, and DEI within the tourism sector by employing SDA. The outcomes of the SDA illuminate the role of supply-driven dynamics—such as the scale and structure of digital supply—in shaping DTV, DTE, and DEI.

By integrating the Ghosh IOA, SA, and SDA approaches, this study presents a comprehensive picture of how digitalization decouples the value-added and carbon emissions of the tourism sector from a supply-driven perspective. These findings offer valuable insights into optimizing value-added while minimizing carbon emissions throughout the digitalization journey in the tourism industry.

2.1. Ghosh Input-Output Analysis

The I-O analysis, grounded in both Leontief and Ghosh models, serves as a fundamental tool for examining economic and environmental interactions among sectors. The traditional Leontief I-O model posits that economic outputs and carbon emissions are propelled by final demand, offering insights into economic and environmental dynamics from a consumption-side perspective. Conversely, the Ghosh I-O model suggests that economic output is stimulated by the supply of primary inputs, framing these activities from a supply-driven viewpoint. Digitalization influences tourism primarily through the provision of digital products and services, with the digitalization of tourism signifying substantial enhancements from digital inputs [29]. Therefore, the Ghosh I-O model is deemed more appropriate for exploring the economic and environmental connections between digitalization and tourism from a supply-driven perspective in this study.

The basic Ghosh I-O model can be expressed as follows:

$$X=u\cdot {\left(I-H\right)}^{-1}=u\cdot G$$

(1)

where $X$ is the matrix that shows the total output of a country; $H$ is the allocation coefficients matrix with elements $H_{ij}=Z_{ij}/X_i$, reflecting the monetary allocation flows from sector $i$ to sector $j$; $I$ is the identity matrix; $u$ is the primary input vector; $G={\left(I-H\right)}^{-1}={\left(G_{ij}\right)}_{n\times n}$ is the Ghosh inverse matrix.

Using the non-competitive imports assumption, the value-added and carbon emissions can be further calculated from Equation (1) as follows:

$$V=X\cdot k=\stackrel{\wedge }{u}\cdot {\left(I-H\right)}^{-1}\cdot k=\stackrel{\wedge }{u}\cdot G\cdot k$$

(2)

$$E=X^{\prime }\cdot f=\stackrel{\wedge }{u}\cdot {\left(I-H\right)}^{-1}\cdot f=\stackrel{\wedge }{u}\cdot G\cdot f$$

(3)

where $V$ and $E$ represent the value-added and carbon emissions driven by primary input, respectively; $u$ denotes primary input vector, $k$ denotes value-added coefficient vector, with elements $k_i=u_i/X_i$ representing the value-added per unit of output produced by sector $i$; f denotes sectoral emission intensity vector, with elements $f_i=e_i/X_i$ and $e_i$ is the direct emissions generated by sector $i$ and ${\left(X_i\right)}_{n\times 1}$ is a vector of sector output.

2.2. Calculation of Digital-Enabled Emission Intensity

Based on the supply-side accounting of DTV and DTE, DEI can be calculated at a sectoral level as follows:

$$DEI={\displaystyle \frac{E}{V}}={\displaystyle \frac{{\displaystyle \sum E^i}}{V}}={\displaystyle \sum \frac{E^i}{V^i}}\cdot {\displaystyle \frac{V^i}{V}}={\displaystyle \sum W^i}\cdot {DEI}^i$$

(4)

where $E$ and $V$ are DTE and DTV, respectively; $DEI$ is digital-enabled emission intensity, indicating carbon performance (or productivity) of tourism caused by one unit of digital primary input, which is defined to analyze the relationship between DTE and DTV from the supply-driven perspective; $w^r$ is the share contributed by the DTV of subsector $i$ and ${\sum }_{i=1}^m{w}^i=1$; ${DEI}^i$ is the digital-enabled intensity at a subsector level, measured as the ratio of DTE to DTV of subsector $i$; However, $DEI$ and ${DEI}^i$ are inverse indicators for measuring carbon performance: the smaller the ratios are (less emissions, or more value-added), the higher the carbon performance is. Therefore, the DEI in tourism can be a weighted sum of tourism subsectors’ DEIs, in which the weight is the ratio of tourism subsectors’ share of DTV in total tourism.

2.3. Subsystem Analysis

To split the economic and environmental impacts of digitalization on tourism from the economy system, the SA approach is applied based on the supply-driven Ghosh I-O model. The I-O table of a country includes many economic sectors. The digital sector and tourism sector can be considered, respectively, as a subsystem of industries. The economic sectors of a country can be classified into three subsystems, i.e., the digital technology subsystem, the tourism subsystem, and the rest of the economy. Following Cheng et al. (2023) and Zhou et al. (2019), the value-added and CO₂ emissions of tourism subsectors enabled by digital technology subsectors can be demonstrated as follows [30,31]:

$$V_{d\to t}=u_d\cdot G_{d\to t}\cdot k_t$$

(5)

$$E_{d\to t}=u_d\cdot G_{d\to t}\cdot f_t$$

(6)

where $V_{d\to t}$ and $E_{d\to t}$ represents the value-added and carbon emission of tourism enabled by digitalization, showing the value-added and carbon emissions generated in tourism services processes induced by the application of digital products; $u_d$ is the primary input vector of the digital industry that contains only the digital sector with zeros elsewhere, showing the total supply of digital products; $G_{d\to t}$ is the Ghosh inverse matrix containing the allocation coefficients for digital sector to tourism but zeroes for all other sectors, presenting the total supply of digital goods and services required to produce one unit products in tourism subsectors (or digital supply structure); $k_t$ is the value-added coefficient vector of tourism industry, $f_t$ is emission intensity vector of tourism industry. Detailed derivation in the Equations (5) and (6) are given in Appendix C.

To inspect the economic and environmental roles played by different digital subsectors and unveil their actions on different tourism subsectors, Equations (7) and (8) can be formulated as

$$S{V}_{d\to t}=\stackrel{\wedge }{u_d}\cdot G_{d\to t}\cdot \stackrel{\wedge }{k_t}$$

(7)

$$S{E}_{d\to t}=\stackrel{\wedge }{u_d}\cdot G_{d\to t}\cdot \stackrel{\wedge }{f_t}$$

(8)

where $S{V}_{d\to t}$ and $S{E}_{d\to t}$ represents the economic flow matrix and the carbon flow matrix, respectively, the row sum of two matrices is the value-added and carbon amount of resource subsectors, and the column sum of two matrices is the value-added and carbon amount of destination subsectors. $\stackrel{\wedge }{u_d}$, $\stackrel{\wedge }{k_t}$, $\stackrel{\wedge }{f_t}$ are the diagonal matrices of $u_d$, $k_t$, $f_t$ respectively.

2.4. Structural Decomposition Analysis

Both the additive and multiplicative SDA can be applied to explore factors that drive the changes in value-added and emissions [32]. In general, the additive SDA is used to explore an absolute change of a quantity indicator while the multiplicative SDA to a relative change of an intensity indicator [33]. We introduce the additive SDA to investigate the drivers of the DTV and DTE and use the multiplicative SDA to analyze the change of the DEI.

With the additive SDA, the vector of digital primary input ($u_d$) in Equations (5) and (6) can be further divided into three components: digital supply scale ($u_{si}$, i.e., total digital primary input), digital supply type structure ($u_{ti}$, i.e., the share of imports, labor, tax, capital, enterprise surplus in total primary input), and digital product structure ($u_{pi}$, i.e., the share of digital products in each primary input type), the absolute change of DTV and DTE from base period 0 to test period T can be decomposed as:

$$\begin{array}{cc}\hfill \Delta V=V^T-V^0=& \left(\Delta u_{si}\right)\cdot u_{ti}\cdot u_{pi}\cdot G\cdot k\hfill \\ & +u_{si}\cdot \left(\Delta u_{ti}\right)\cdot u_{pi}\cdot G\cdot k\hfill \\ & +u_{si}\cdot u_{ti}\cdot \left(\Delta u_{pi}\right)\cdot G\cdot k\hfill \\ & +u_{si}\cdot u_{ti}\cdot u_{pi}\cdot \left(\Delta G\right)\cdot k\hfill \\ & +u_{si}\cdot u_{ti}\cdot u_{pi}\cdot G\cdot \left(\Delta k\right)\hfill \end{array}$$

(9)

$$\begin{array}{cc}\hfill \Delta E=E^T-E^0=& \left(\Delta u_{si}\right)\cdot u_{ti}\cdot u_{pi}\cdot G\cdot f\hfill \\ & +u_{si}\cdot \left(\Delta u_{ti}\right)\cdot u_{pi}\cdot G\cdot f\hfill \\ & +u_{si}\cdot u_{ti}\cdot ;\left(\Delta u_{pi}\right)\cdot G\cdot f\hfill \\ & +u_{si}\cdot u_{ti}\cdot u_{pi}\cdot \left(\Delta G\right)\cdot f\hfill \\ & +u_{si}\cdot u_{ti}\cdot u_{pi}\cdot G\cdot \left(\Delta f\right)\hfill \end{array}$$

(10)

where $\Delta$ denotes changes of variables; here $\Delta V$ can be decomposed into changes in five factors, i.e., the effects of digital supply level, supply type structure, digital product structure, digital application structure, sectoral value-added coefficient; similarly,$\Delta E$ can be decomposed into changes in five factors, i.e., the effects of digital supply level, digital supply type structure, digital product structure, digital application structure, and sectoral emission intensity. They display the contributions of each driving factors to the changes of the value-added and carbon emissions in tourism enabled by digital technology. We use the D&L method here to derive the contributions of these driving factors [34].

With the multiplicative SDA, the relative change of DEI in tourism from base period 0 to test period T can be measured as follows:

$$\begin{array}{cc}\hfill D_{tot}& ={\displaystyle \frac{DC{I}^t}{DC{I}^0}}={\displaystyle \frac{E^t/V^t}{E^0/V^0}}={\displaystyle \frac{E^t/E^0}{V^t/V^0}}\hfill \\ & ={\displaystyle \frac{D_{E,us}{D}_{E,ut}{D}_{E,up}{D}_{E,G}{D}_{E,f}}{D_{V,us}{D}_{V,ut}{D}_{V,up}{D}_{V,G}{D}_{V,k}}}\hfill \\ & =D_{tot,us}\times D_{tot,ut}\times D_{tot,up}\times D_{tot,G}\times D_{tot,f}\times D_{tot,k}\hfill \end{array}$$

(11)

where $D_{tot,us}$, $D_{tot,ut}$, $D_{tot,up}$, $D_{tot,G}$, $D_{tot,f}$ and $D_{tot,k}$ are the six sub-effects driving emission intensity change; $D_{tot,us}$ is the primary input scale effect, showing the contribution of changes in digital primary input scale, $D_{tot,ut}$ is the primary input structure effect, showing the contribution of changes in digital primary input structure; $D_{tot,up}$ is the primary input product effect, showing the contribution of changes in digital primary input product; $D_{tot,G}$ is the application structure effect, showing the contribution of intermediate application structure change of digital sector-to-tourism sector; $D_{tot,f}$ is the carbon efficiency effect, showing the contribution of tourism emission intensity change; $D_{tot,k}$ is the value-added coefficient, showing the contribution of value-added coefficient change in tourism. If $D_{tot}>1$, the larger the value of $D_{tot}$ is, the greater the increasing effect on $DEI$ is; if $D_{tot}<1$, the smaller the value of $D_{tot}$ is, the greater the decreasing effect on $DEI$ is. We identify the major factors driving changes in $DEI$ using the D&L method as well, which can effectively balance the precision of decomposition results and computational complicacy in the case of a relatively small number of factors [35].

2.5. Data Treatment

Assessing the economic and carbon impacts of digitalization on tourism within an I-O framework necessitates three distinct datasets: monetary I-O tables, tourism economic activity data, and sectoral CO₂ emissions inventory data. National I-O tables for China from the years 2002, 2007, 2012, and 2017 were obtained from the National Bureau of Statistics [36,37,38,39]. These tables, featuring varying sectoral classifications, were harmonized into a consistent framework of 100 sector divisions. This harmonization includes seven digital subsectors, as defined by China’s Statistical Classification of Digital Economy and its Core Industries [40], seven tourism subsectors as per China’s National Statistical Classification of Tourism and Related Industries [27], and 86 other sectors aligned with the most recent National Standard for Industrial Classification for Economic Activities of China [41].

China’s national I-O tables are compiled under the competitive imports assumption and presented in current prices. To align with the non-competitive import assumption of the Ghosh I-O model, the intermediate and final demand of imports are excluded from the four I-O tables, assuming a consistent import ratio across intermediate inputs and final demands. Subsequently, the 2007, 2012, and 2017 I-O tables were adjusted to 2002 constant prices using the double-deflation method. Sector-level price indices were sourced from the National Bureau of Statistics, employing producer price indices for deflating the agricultural and industrial sectors, fixed asset investment price indices for construction, and consumer price indices for service sectors. Specifically, for digital subsectors, electronic equipment indices were used for digital equipment subsectors, and communication service indices for digital services subsectors. Given that tourism subsectors fall under service sectors, consumer price indices were applied for deflation. These deflators, encompassing price indices for 49 sectors, were subsequently applied to the 100 sectors represented in the I-O tables.

Tourism economic activity data were collected from the Survey of China’s Domestic Tourism Consumption [42,43,44,45]. In this survey, the proportion of tourism output within the total output of tourism-oriented sectors is determined by dividing the receipts from tourism subsectors by the total output of related subsectors in the corresponding years of I-O tables. Subsequently, tourism subsectors are delineated from the broader tourism-oriented sectors in the I-O tables by multiplying the rows representing tourism-oriented sectors by the calculated share of tourism output. This process is detailed in Appendix D.

China’s sectoral CO₂ emissions inventory for the respective years was sourced from the China CO₂ Emission Accounts [46,47]. This inventory details the direct CO₂ emissions from 47 sectors within China, adhering to the guidelines set forth by the IPCC. To estimate the direct emissions of tourism subsectors, for which direct emissions and energy consumption data were not explicitly available in statistical records, we employed the output weight-shares treatment approach. This method assumes that the emission intensity of tourism subsectors aligns with that of their corresponding sectors as originally documented.

3. Results and Discussion

3.1. DTV and DTE

We initially calculated the contributions of digitization to the value-added and carbon emissions in tourism using Equations (5) and (6), discovering significant increases in both metrics from 2002 to 2017 (Figure 2). Notably, the growth rate of value-added surpassed that of carbon emissions. Specifically, between 2002 and 2017, digitization’s contribution to tourism’s value-added escalated by 18.3 times, from CNY4.37 billion to CNY78.89 billion (Figure 2a), whereas its emissions contribution rose by 11.3 times, from 0.38 to 4.3 megatons (Mt) of CO₂ (Figure 2b). However, DTV dropped off to CNY31.09 billion while DVE declined to 1.57 Mt in 2020, due to the COVID-19 epidemic.

Although the overarching trends align, the contributions of digitization to value-added and emissions across tourism subsectors show marked disparities. Notably, the share of tourism transportation in value-added witnessed a substantial rise from 19% in 2002 to 44% in 2017, before falling back to 36.18% in 2020 (Figure 2c). Concurrently, tourism transportation emerged as the predominant source of emissions, with its contribution escalating from 51.32% in 2002 to 80.45% in 2017, and slightly increasing to 81.21% in 2020 (Figure 2d). This shift can be primarily attributed to tourism transportation’s relatively high emission intensity compared to other subsectors, compounded by a shift in travel preferences from agency-coordinated bus tours to convenient independent air travel, especially for overnight trips by individual travelers. Conversely, comprehensive services saw a continuous decline in their shares, including during the COVID-19 pandemic, reflecting a shift from traditional travel agency bookings to independent travel facilitated by advancements in digital technology.

From 2002 to 2017, the significant contributors to DTV were primarily tourism transportation, shopping, accommodation, and comprehensive services. In 2017, the digital supply catering to the production needs of these subsectors amounted to CNY34.9 billion for tourism transportation, CNY12.9 billion for tourism shopping, CNY12.4 billion for tourism accommodation, and CNY8.26 billion for comprehensive services, collectively comprising over 85.7% of the total DTV. Moreover, the supply of digitalization facilitated approximately 11.2% of the DTV in the tourism catering and sightseeing sectors. Conversely, tourism entertainment contributed minimally to the value-added. However, in 2020, due to the sharp decrease in the number of tourists caused by the COVID-19 pandemic, the digital supply attending to the production needs of these subsectors shrank to CNY11.2 billion for tourism transportation, CNY9.5 billion for tourism shopping, CNY4.8 billion for tourism accommodation, and CNY0.97 billion for comprehensive services.

From 2002 to 2017, the distribution of DTV within tourism subsectors showed a trend toward greater concentration. Specifically, the proportion of DTV attributed to tourism transportation saw a substantial rise, growing from 19% in 2002 to 44% in 2017. Conversely, the share allocated to comprehensive services experienced a notable decline, dropping from 33% in 2002 to 10% in 2017, and further narrowing to 3.12% in 2020. Additionally, there was an increase in the DTV derived from tourism accommodation and sightseeing during this period. During the COVID-19 pandemic, the shares of most tourism subsectors decreased, except for tourism shopping and tourism catering.

The distribution of DTE mirrors similar patterns from 2002 to 2017. The emissions resulting from the supply of digitization predominantly occurred in subsectors such as tourism transportation, comprehensive services, tourism accommodation, and tourism shopping. In 2017, the supply of digitization led these tourism subsectors to emit 3.46, 0.33, 0.15, and 0.14 Mt of CO₂, respectively, representing 95% of the total DTE in the tourism sector. Notably, the leading subsectors contributing to DTE closely align with those contributing to DTV. Regarding shifts in distribution among tourism subsectors, the proportion of DTE attributed to tourism transportation surged by 29% from 2002 to 2017. Conversely, the shares allocated to comprehensive services, shopping, and accommodation experienced declines of 23%, 3.9%, and 1.7%, respectively. Meanwhile, the contributions from tourism catering, sightseeing, and entertainment exhibited minor fluctuations throughout the same period. From 2017 to 2020, the share of DTE attributed to tourism shopping increased by 3.68%, while that of comprehensive services decreased by 4.46%; the shares of other subsectors fluctuated slightly.

3.2. Decoupling Status and DEI

To evaluate the variations in the rate of change in value-added and carbon emissions in the tourism sector enabled by digitization, the Tapio Decoupling Index (TDI) is employed. This index calculates the percentage rise in DTE of tourism for every percentage increase in DTV, offering a precise metric of the balance between economic gains and environmental effects [28]. For the tourism sector, both DTV and DTE have increased, indicating a weak decoupling of its total digital-enabled value-added from carbon emissions between 2002 and 2020 (TDI = 0.44). The analysis period is segmented into four phases: Period 1 (2002–2007), Period 2 (2007–2012), Period 3 (2012–2017), and Period 4 (2017–2020), as shown in Figure 3.

In the initial period (2002–2007), the sector experienced expansive coupling between DTV and DTE (TDI = 1.12). Tourism accommodation demonstrated weak decoupling, with increased value-added and decreased emissions. Conversely, comprehensive services and catering, with a carbon emission rate slightly exceeding the value-added growth rate, exhibited expansive coupling. The tourism transportation sector approached the boundary between expansive coupling and weak decoupling. Other tourism subsectors were characterized by weak decoupling.

From 2007 to 2012, the sector transitioned from expansive coupling to weak decoupling (TDI = 0.48), with comprehensive services achieving strong decoupling and tourism sightseeing approaching strong decoupling. Comprehensive services, previously exhibiting expansive coupling, shifted to strong decoupling, with a lower rate of emission increase and a higher rate of value-added growth than in the preceding period. This period saw all subsectors either strongly or weakly decoupled in terms of DTV and DTE.

From 2012 to 2017, the sector maintained weak decoupling (TDI = 0.68), albeit with a tendency towards the higher end of weak decoupling compared to the previous period. Comprehensive services remained strongly decoupled (TDI = −0.81). Meanwhile, tourism catering exhibited expansive coupling (TDI = 1.25), with economic growth and carbon emissions changing at a similar rate due to a higher increase in emissions. This suggests a relatively unstable decoupling relationship in tourism catering, shifting from weak decoupling to expansive coupling between the second and third periods.

In the final period (2017–2020), the sector showed recessive coupling (TDI = 1.04), with decreased emissions and declining value-added. Tourism catering demonstrated recessive decoupling (TDI = 2.3), with economic growth and carbon emissions declining and a higher decrease in emissions. All other subsectors showed recessive coupling in terms of DTV and DTE. This is largely due to the sudden decrease in consumption caused by the COVID-19 pandemic. Driven by digitization, the tourism sector has experienced significant economic growth while also generating considerable emissions. Nonetheless, there is notable subsectoral variation in the rates of change between value-added and carbon emissions. Comprehensive services, with their declining emissions and rising value-added, play a pivotal role in fostering a DTV. It is noteworthy that, despite bearing the brunt of DTE, tourism transportation’s value-added growth rate exceeds its emission increase rate. This suggests that while DTE of transportation may rise with its value-added growth, the latter is expected to outpace the former, indicative of weak decoupling between economic benefits and carbon emissions. Such decoupling is crucial for the sustainable development of the tourism sector.

To assess the dual contributions of digitization to tourism—emissions and value-added—the concept of DEI was calculated using Equation (4). This metric quantifies the DTE per unit of DTV originating from the digital sector, thus effectively reflects the environmental costs against the economic benefits prompted by digital supply. As shown in Figure 4, both the tourism sector and its subsectors exhibited declining DEIs from 2002 to 2020, signifying an enhancement in their emission efficiency.

Regarding subsectors, a uniform decrease in DEI was observed during the same period, indicating that all subsectors experienced a significant reduction in DEI. However, notable differences exist among them. Despite achieving the largest reduction in DEI, tourism transportation remains the subsector with the highest DEI, attributed to its substantial direct emission intensity. This disparity highlights the considerable room for improvement, particularly emphasizing the need for more rigorous energy efficiency measures in tourism transportation. The DEI reduction trajectory of comprehensive services closely mirrors that of the total tourism sector, while DEIs of other subsectors are consistently lower than that of the overall sector throughout 2002–2020.

The subsectoral contributions of DEI varied significantly between 2002 and 2020, exhibiting considerable fluctuations (Figure 5). As outlined in Equation (4), the total tourism DEI can be broken down into a weighted sum of subsectoral DEIs. For comprehensive services, the period 2002–2007 saw a positive contribution of 23%, with its share increasing from 33% in 2002 to 47.8% in 2007. However, this trend reversed between 2007 and 2012, where it had the largest negative impact (35%), and its share decreased to 23.7% in 2012. Conversely, tourism transportation’s narrative was opposite during 2002–2012; it had a negative contribution of −16% from 2002 to 2007, with its share decreasing from 19.3% in 2002 to 16.95% in 2007. This changed positively between 2007 and 2012 when it made the largest positive contribution (34%), and its share rose to 31.4% in 2012.

This fluctuation between comprehensive services and tourism transportation reflects a shifting preference among tourists between group and independent travel, influenced by the widespread adoption of digital technology in tourism and the popularity of private vehicles, as indicated by the Survey of China’s Domestic Tourism Consumption [21,42,43,44,45]. The effects—both positive from tourism transportation and negative from comprehensive services—narrowed in the period 2012–2017 and 2017–2020 compared to 2007–2012, demonstrating a reduction in emission intensity and an optimization in the allocation of digital technology for these subsectors.

Except for tourism transportation and, to a minor extent, tourism catering, all other tourism subsectors contributed negatively to tourism DEIs over the study period between 2002 and 2017. This pattern suggests that either an increase in weight or a reduction in the DEI by subsectors could lead to a more substantial decrease in the overall tourism DEI. Tourism transportation and comprehensive services, given their significant shares of DTV, had a larger influence on DEI due to the impact of the same magnitude of emission intensity decline. This also indicates a broader trend of decreasing emission intensity across subsectors and optimizing the digital technology allocation structure, leading to a diminished positive effect from tourism transportation and a prevailing negative contribution from other subsectors (except for a slight increase from tourism catering). In the period 2017–2020, the share of transportation has decreased from 43.7% to 36%, while the share of shopping has increased from 16.2% to 30.6%. The contribution of transportation carbon intensity has only increased by 1%, while shopping has shifted from negative to increasing by 4%.

3.3. Transfer Flows of DTV and DTE

We conducted a subsystem analysis to pinpoint the primary digital subsectors contributing to both value-added and carbon emissions within tourism subsectors and to determine the main tourism subsectors that receive value-added and carbon emissions from digital subsectors. Employing Equations (7) and (8), we traced and mapped the flows of value-added and emissions from digital subsectors to tourism subsectors in 2017 and 2020, as depicted in Figure 6 and Figure 7. These figures effectively illustrate the transfer of value-added and carbon emissions between the digital sector, serving as the source sector, and the tourism sector, acting as the destination sector, highlighting the net forward perspective.

From the perspective of value-added flow sources, three digital subsectors account for a significant portion of the value-added transfer, namely software and IT services, electronic components, and communication services, contributing CNY25.62 billion, CNY16.74 billion, and CNY15.09 billion to tourism in 2017, respectively. Among these, software and IT services emerged as the most substantial source of value-added, channeling 32.1% of the total value-added to various tourism subsectors. Hence, enhancing primary inputs in software and IT services plays a crucial role in amplifying tourism’s value-added. Electronic components and communication services stood as the next two pivotal source subsectors, collectively transferring 40% of the value-added to tourism subsectors. Enhancing the efficacy of these digital subsectors represents an effective strategy to increase tourism value-added. In 2020, software and IT services, electronic components, and communication services still contributed the majority of the value-added transfer; however, they showed different trends in change. The share of software and IT services and electronic components continued to grow to 37% and 26%, respectively, while that of communication services decreased to 14%.

Viewed from the destination perspective of value-added flow, tourism transportation absorbed 44% of the value-added originating from digital subsectors in 2017. The value-added flow was significantly more pronounced between tourism accommodation, tourism shopping, and the digital sector compared to tourism catering, entertainment, and comprehensive services. This discrepancy arises not only because the operations of tourism accommodation and shopping services necessitate digital equipment and services but also due to the heavy reliance on online reservation systems for tourism accommodation on digital services. In contrast, tourism entertainment and catering exhibit a lesser dependence on the digital industry. In 2020, the share of tourism transportation decreased to 36%, yet it remained the largest part of DTV, and tourism shopping increased to 30.6% with an increased growth rate. Tourism catering is another subsector that showed growth in DTV share, increasing to 10%, while that of other subsectors showed varying degrees of decline.

From the perspective of emission flow sources, digital services (i.e., information services and communication services) are the predominant contributors to emissions. In 2017, they accounted for 2.28 Mt of CO₂, representing approximately 53.1% of total DTE. Notably, software and IT services were the largest emitters, generating 1.47 Mt of CO₂ and constituting 34.2% of the DTE. Compared to digital services, emissions from digital equipment were lower, with electronic components being the main source. This disparity arises partly due to variances in digital primary inputs; for example, in 2017, the primary input for digital services was CNY1200 billion, significantly higher than the CNY100 billion inputs for computers, audio-visual apparatus, and other communication and electronic equipment. Additionally, digital subsectors impact tourism subsectors’ production activities differently. For instance, communication equipment and audio-visual apparatus resulted in lower emissions compared to other communication and electronic equipment, despite having higher primary inputs. In 2020, software and IT services accounted for 0.7 Mt of CO₂, increasing to 44% of total DTE, while electronic components and communication services declined to 15% and 16%, respectively.

From the perspective of emission flow destinations, digital supply predominantly impacts tourism transportation, which was responsible for producing 3.46 Mt CO₂ in 2017, making up 80.5% of total DTE. This finding aligns with Zhou et al. (2022), who observed that the application of digital products tends to increase emissions in carbon-intensive sectors [19]. Furthermore, digital supply contributed over 0.76 Mt of CO₂ across tourism shopping, accommodation, catering, and comprehensive services, with the least impact on tourism sightseeing and entertainment emissions. Therefore, compared to other tourism subsectors, decarbonizing tourism transportation emerges as a critical priority. In 2020, the carbon emissions of the tourism transportation subsector were 1.28 Mt CO₂, with a slight increase in proportion (0.7% increase), and a decrease of 8% in its DTV proportion. Similar to the change in DTV, the contribution of tourism shopping to DTE increased to 7%, while that of comprehensive tourism services decreased to 3%. To examine the evolving flow between 2002 and 2020, we illustrate the comprehensive value-added and emissions transferred from digital subsectors to the tourism sector as the destination in Figure 8.

From the perspective of value-added flow sources, digital subsectors have channeled increasingly significant amounts of value-added to tourism subsectors during 2002–2017, with DTV exhibiting a trend towards greater concentration. The contribution from communication services to DTV decreased dramatically, from 55% in 2002 to 18% in 2017. Similarly, the computer subsector saw its share diminish from 19% in 2002 to 10% in 2017. In contrast, the share of software and IT services surged remarkably from 10% to 32%, marking the most substantial growth in value-added transfer to tourism subsectors. Furthermore, electronic components and other communication and electronic equipment have increased their contributions, adding 12.8% and 8.6% to the DTV share, respectively, which warrants attention. The shares of communication equipment and audio-visual apparatus experienced minimal changes. During 2017–2020, the proportion of software and IT services in DTA increased from 32% in 2017 to 38% in 2020, while that of communication services decreased from 19% to 14%. Examining the reasons, the primary inputs in software and IT services and communication services increased by 90.7 billion and 12 billion, respectively, while those of other subsectors decreased to varying degrees. Consequently, the proportion of software and IT services in the total value of digital products in the tourism sector increased from 31% to 41%, whereas the proportion of communication services decreased from 11% to 8%. The proportion of electronic components in DTA has increased from 21% in 2017 to 26% in 2020, despite a decrease in the primary input of electronic components.

From the perspective of emission flow sources, the narrative mirrors that of value-added. The proportion of DTE originating from communication services fell by 39.2%, slightly more than its value-added decrease of 36.7% between 2002 and 2017. Conversely, the share of software and IT services in emissions rose by 25.4%, marginally exceeding its value-added increase of 22%. Electronic components and other communication and electronic equipment saw their shares increase by 13% and 9.3%, respectively, while the computer subsector experienced a 10.1% decline. Communication equipment and audio-visual apparatus showed negligible fluctuations during this period. Notably, software and IT services have emerged as a pivotal subsector contributing to a significant rise in carbon emissions to tourism, particularly post-2012, thereby highlighting its critical role in emission reduction efforts. During 2017–2020, the total DTE decreased by 2.72 million tons, with all subsectors experiencing varying degrees of decline. As for the contribution share of subsectors, the proportion of software and IT services in DTE increased from 34% in 2017 to 44%, while the shares of computers, electronic components, and audio-visual apparatus decreased by 2%, 6%, and 3%, respectively. Communication equipment showed little fluctuation. It is worth mentioning that electronic components demonstrated good carbon performance, with increasing DTA share and decreasing DTE share. Therefore, policymakers should closely monitor the enabling impact of software and IT services alongside electronic components. As these subsectors predominantly function as intermediate goods for tourism subsectors, they play crucial roles in augmenting value-added, which consequently leads to higher carbon emissions.

To examine the evolving flow of sectors between 2002 and 2020, we illustrate the comprehensive value-added and emissions transferred from the digital sector as the source to various tourism destination subsectors in Figure 9.

For value-added changes, all tourism subsectors experienced a significant increase, especially tourism transportation, which witnessed the most remarkable growth from CNY0.8 billion in 2002 to CNY35 billion in 2017. Regarding emissions, there was a noticeable trend towards concentration in DTE; tourism transportation experienced a substantial rise from 0.2 Mt in 2002 to 3.5 Mt in 2017. In contrast, other tourism subsectors exhibited minor fluctuations. During 2017–2020, DTV shrank by 61.1%, but the tourism subsectors did not decrease proportionally. The most significant changes appeared in tourism transportation and tourism shopping. Compared with 2017, the share of tourism transportation decreased by 7.5%, falling back to CNY11.25 billion in 2020, while the share of tourism shopping increased by 14.47%, decreasing to CNY9.5 billion in 2020.

The proportion of tourism transportation in DTE escalated from 54% in 2002 to 80.5% in 2017, and its share in DTV rose from 19% in 2002 to 44% in 2017, indicating a parallel increase in both value-added and emissions. However, the proportion of tourism transportation in DTE continuously rose to 81.2% while that in DTV shrank to 36% in 2020, showing an imbalance in development between value-added and emissions. This is probably because the number of passengers decreased while the fuel required for transportation did not decrease proportionally. Conversely, the share of comprehensive services in DTE plummeted from 28% in 2002 to 7% in 2017, continuously declining to 3.3%, and its contribution to DTV dropped from 33% in 2002 to 10% in 2017, further down to 3.1% in 2020, showing a simultaneous decline in both metrics. On the other hand, the share of tourism accommodation in DTV ascended from 13.4% in 2002 to 15.6% in 2017, maintaining 15.4% in 2020, whereas its proportion in DTE decreased from 5.2% in 2002 to 3.5% in 2017, slightly rising to 3.7% in 2020, demonstrating a net gain. The disparity in trend changes among destination sectors underscores the varied adoption and diffusion rates of digital technology across tourism subsectors in China. Consequently, there is a pressing need to enact policies that enhance value-added and reduce emissions tailored to the distinctive development needs and emission profiles of tourism subsectors, leveraging digital technology effectively.

Throughout the study period, digitization consistently and significantly influenced tourism subsectors, revealing sectoral heterogeneity in its economic and carbon impacts. While software and IT services significantly boosted both value-added and carbon emissions, the contributions from communication services decreased by nearly two-thirds. In terms of destination subsectors, digitization notably enhanced value-added in tourism shopping, transportation, and accommodation, yet it also concentrated the majority of carbon emissions in the transportation subsector. Therefore, strategically optimizing the structure of digital supply could be key to enhancing value-added and mitigating carbon emissions within the tourism industry.

3.4. Drivers of DTV, DTE, and DEI

To delve deeper into the mechanisms behind the changes, we utilized SDA to pinpoint the primary drivers behind the changes in DTV, DTE, and DEI. Using Equation (9), we broke down the shifts in DTV on tourism between 2002 and 2017 into five key drivers: the scale of digital supply, the structure of digital types, the structure of digital products, the structure of digital applications, and the value-added coefficient of tourism. Figure 10 illustrates the impact of these factors on the digital-enabled economic contributions to tourism across various time frames.

DTV experienced a substantial increase of CNY75.52 billion from 2002 to 2017, expanding by an 18.3-fold growth rate. This surge was primarily driven by the expansion of the digital supply level, which escalated from CNY0.18 billion to CNY2.03 billion, contributing CNY57 billion to the value-added growth. Moreover, alterations in the digital application structure significantly bolstered economic growth, adding an effect of CNY23.24 billion during 2002–2017. The change in the value-added coefficient of tourism had a smaller impact of CNY0.18 billion. Conversely, modifications in the digital supply type structure (including labor compensation, production taxes, fixed assets depreciation, and operating surplus) and digital product structure slightly hindered DTV growth, detracting CNY2.63 billion and CNY7.3 billion, respectively. These findings highlight the substantial improvements in digital supply level, digital application structure, and the value-added coefficient of tourism during 2002–2017.

Focusing on growth intervals from 2002 to 2017, the latter half of the growth in DTV (66%) materialized between 2012 and 2017, with slower growth rates observed in the preceding intervals, contributing 12.1% and 21.9% to the total increase, respectively. This shift can be attributed to the sustained increase in digital supply scale across each sub-period and the transformation of digital application structure impacts from negative to positive post-2012. The impact of digital application structure changes decreased by CNY5.13 billion from 2002 to 2007 but climbed by CNY3.6 billion from 2007 to 2012 and surged by CNY24.76 billion from 2012 to 2017. Meanwhile, the value-added coefficient of tourism saw increases of CNY0.66 billion from 2002 to 2007 and CNY0.33 billion from 2007 to 2012 but experienced a decrease of CNY0.82 billion from 2012 to 2017, still resulting in an incremental overall effect. These dynamics reflect significant structural shifts in digital application around 2012, suggesting that a combination of digital supply level and digital application structure advancements would continue to accelerate DTV growth.

However, DTV decreased by half from 2017 to 2020, dropping to CNY31.09 billion, slightly higher than the level in 2012. Over this period, the digital supply level increased by CNY0.59 billion, contributing CNY14.6 billion to the value-added growth in tourism. Digital product structure added an effect of CNY4.48 billion for DTV growth. In contrast, the digital application structure significantly undermined economic growth, subtracting an effect of CNY52.3 billion during 2017–2020. The value-added coefficient of tourism and the digital supply type structure hampered DTV growth, detracting CNY14.9 billion and CNY0.61 billion, respectively.

On the subsector level, increases in the digital supply scale of software and IT services and electronic components were the largest contributors to DTV growth in 2017, accounting for 54.5% of the increasing effects of the digital supply scale with rises of CNY13.5 billion and CNY17.5 billion, respectively. This was followed by communication equipment, computers, communication services, and other digital subsectors, which also contributed to economic growth in tourism through enhanced digital application structures. The expansion of digital supply scale directly facilitated the growth of value-added in tourism.

Similarly, using Equation (10), the changes in DTE for the same period were dissected into five driving factors: the scale of digital supply, the structure of digital types, the structure of digital products, the structure of digital applications, and the emission intensity of tourism subsectors. Figure 11 outlines how these elements influenced DTE over the study periods.

DTE witnessed a rapid increase, rising by 3.92 Mt CO₂ and achieving an 11-fold growth from 2002 to 2017, as depicted in Figure 11. The expansion of the digital supply scale emerged as the predominant factor driving the surge in DTE, it resulted in a 4.1 Mt increase in DTE. The digital application structure ranked as the second most influential driver, contributing to a 2.1 Mt growth in DTE, while the contribution of DTV rose from CNY−5.13 billion in 2002 to CNY24.76 billion in 2017, marking a significant rise in its proportional impact. Conversely, changes in digital type structure—such as the proportions of imports, labor, capital, enterprise surplus, and taxes—and digital product structure exerted slight inhibitory effects of −0.13 and −0.23 Mt CO₂, respectively. The subdivision of the digital primary input structure had minimal impact on DTE. Notably, the carbon intensity of tourism showcased the most substantial inhibitory effect, reducing DTE by 1.84 Mt CO₂, indicating a significant improvement in carbon efficiency among downstream users of digital products in China.

The primary increase in DTE predominantly occurred between 2012 and 2017, accounting for 58% of the total increase. This surge can be attributed to the enhanced growth of the digital supply scale and the considerable emission increase driven by the digital application structure. Although reductions in tourism emission intensity continued during 2002–2017, the rate of decrease lagged behind the growth of the digital supply scale. Thus, emission reduction efforts should target digital supply, application structure, and tourism emission intensity to fully exploit their potential for emission mitigation. During 2017–2020, DTE showed a sharp drop, decreasing by 63.3%. The digital supply scale still contributed an increase of 0.77 Mt CO₂, while the digital application structure and tourism emission intensity had inhibitory effects, declining by 2.97 Mt CO₂ and 0.8 Mt CO₂, respectively.

Subsector analysis revealed that the reduction in tourism emission intensity was the most significant contributor to the decline in DTE from 2002 to 2017, with tourism emission intensity accounting for 47% of the total reduction effect. In particular, tourism transportation and accommodation, which have high proportions of digital applications, were significantly impacted by digital supply, leading to substantial carbon emissions.

Between 2007 and 2012, changes in digital application in these two subsectors resulted in increases of 0.19 and 0.17 Mt CO₂, respectively, constituting 76.5% of the emission increase effects from digital application structure. From 2012 to 2017, the supply and application of digital technology, especially in communication services and software and IT services, to these subsectors surged, contributing to 69% of the emission increase effect from digital application structure.

In summary, the dynamics of DTE reflect a tension between the growth of the digital supply scale and the enhancement of carbon efficiency in tourism. To mitigate this, energy conservation, emission reduction in tourism, and judicious digital investment in relevant subsectors should be further promoted.

To analyze the contributor of DEI changes from an environmental-economic viewpoint, Equation (11) is employed to dissect the DEI changes into six sub-effects: the scale effect of digital primary input, the structure effect of digital primary input, the product effect of digital primary input, the effect of digital application structure, the carbon efficiency effect in tourism, and the effect of the value-added coefficient in tourism. Figure 12 presents the decomposition results, illustrating how each factor contributed to the changes in DEI.

From 2002 to 2020, China’s DEI notably decreased from 0.087 to 0.05 kg-CO₂/RMB, representing a reduction rate of 42%. This trend aligns with prior research indicating a negative correlation between digital investment and China’s energy intensity [48]. Among the six identified factors, the carbon efficiency factor was the predominant contributor to the DEI reduction during this period, significantly surpassing the impact of other factors. Conversely, changes in the digital application structure partially neutralized this positive effect. These findings suggest that China’s primary strategies for lowering emission intensity in tourism have centered on promoting energy conservation and adopting cleaner production technologies, alongside leveraging digitalization to enhance tourism revenue and production efficiency. Nonetheless, the potential of optimizing the digital application structure to further reduce emission intensity has received less attention, despite its promise for more cost-effective emission intensity reductions.

The impact of value-added coefficient changes on DEI transitioned from a reducing to an enhancing effect throughout the study period, emerging as a pivotal element in the rise of DEI. Initially, this factor contributed to a decrease in DEI from 2002 to 2012; however, it became a critical driver of DEI growth from 2012 to 2020. The digital primary input product factor displayed a similar pattern, initially suppressing DEI between 2002 and 2007, and then contributing to an increase in DEI from 2007 to 2020. Despite this, the early inhibitory effects of both the value-added coefficient and primary input product on DEI were slightly lower than their later facilitative impacts, resulting in an overall marginal increasing effect on DEI growth. The persistent and increasing influence of the tourism value-added coefficient on DEI stood out as the most substantial factor, marking it as the dominant driver of DEI elevation from 2002 to 2020. A similar pattern was observed for the application structure from 2002 to 2017. This indicates that future efforts to decrease DEI should focus on promoting the tourism value-added coefficient and optimizing the application structure, underscoring their critical role in achieving further reductions in DEI.

3.5. Sensitivity Analysis

The findings of this study are subject to several sources of uncertainty arising from sector aggregation, uniform assumptions of sectoral emission intensities, consistent import ratios, and the chosen study period.

Due to data limitations, we follow Su and Ang (2013) and Zhou et al. (2022) by aggregating specific sectors in the national input-output table to align more effectively with our carbon-emission dataset [19,49]. This aggregation approach presumes identical emission intensities and value-added levels across merged subsectors. Specifically, we consolidated tourism subsectors including tourism transportation, tourism entertainment, and comprehensive services (see Table A2 in Appendix B). Merging tourism entertainment and comprehensive services is unlikely to significantly affect implicit carbon emissions and value-added estimates, as both subsectors fall within the broader services category and share similar production technologies and operational practices.

In contrast, the tourism transportation sector combines passenger-related portions of multiple transport modes—railway, road, water, air, and urban public transportation—into one aggregate category. This consolidation potentially influences implicit carbon emissions, implicit value-added, and the subsequent decoupling analysis significantly, due to notable operational and technological differences among these transport modes. Assuming uniform carbon emission intensities and value-added levels may skew the decoupling outcomes of the tourism subsector toward the overall industry average. Future research, contingent upon availability of detailed, mode-specific emissions data, would allow for a more precise decoupling analysis across distinct tourism travel modes.

Furthermore, when transforming the competitive input-output table into its non-competitive form, we assumed a uniform import ratio for intermediate inputs and final demand. However, this assumption is unlikely to materially distort implicit carbon emissions and value-added calculations or affect our decoupling conclusions, given the tourism services sector’s relatively homogeneous production processes and involves relatively little processing and re-export trade [49]. Intermediate inputs and final demand exhibit similar carbon intensities and value-added structures within this sector.

Lastly, our examination of digitalization’s role in the decoupling between the tourism economy and carbon emissions encompasses the COVID-19 period. Although tourism activity drastically decreased globally during the pandemic, resulting in temporary emission reductions, this decline proved unsustainable as emissions rebounded rapidly following economic recovery. Nevertheless, we highlight structural changes during this recessionary coupling phase—particularly, lasting digital technology adoption—that support a low-carbon transition in tourism. Intelligent Transportation Systems (ITS), for instance, facilitate route optimization, reducing unnecessary emissions. Similarly, AI-powered visitor-flow predictions integrated with ticketing and shuttle scheduling help mitigate energy wastage from overcrowding; a notable example is the Palace Museum’s crowd-control system, which achieved a 15 percent reduction in air-conditioning and lighting emissions. Consequently, we anticipate our core findings to remain robust over the foreseeable future.

4. Conclusions and Policy Implications

4.1. Conclusions

Since the turn of the millennium, digitization has been progressively embedded within the tourism industry, making the evaluation of its impacts imperative for fostering sustainable development objectives. This study proposes an integrated input-output analytical framework to examine how digitalization contributes to decoupling tourism economy from carbon emissions. The results reveal that digitization significantly propels tourism economic expansion while moderating its carbon footprint growth. Consequently, digitalization has been instrumental in decoupling economic gains from carbon emissions in the tourism sector. The empirical results of this study support the following conclusions.

First, the surge in value-added within the tourism sector, driven by digitalization, has increased by an astounding 18 times from 2002 to 2017, markedly outpacing the 11-fold increase in carbon emissions associated with digitalization processes within the sector. The economic benefits derived from digitalization have outweighed the environmental impacts. Simultaneously, the decoupling index within the tourism sector has shifted from expansive coupling to weak decoupling throughout the analysis period, highlighting digitalization’s pivotal contribution to the tourism sector’s sustainable development from the perspective of supply-driven dynamics.

Second, significant variations exist in the contributions to value-added and carbon emissions among the tourism and digital subsectors. Tourism transportation has been identified as the primary contributor to both economic gains and carbon emissions within the tourism industry. This situation mirrors a broader trend within the digital sector—the engine behind digitalization—where software and IT services, electronic components, and communication services have significantly propelled DTV. The synergy between digital subsectors, such as software and IT services, electronic components, and certain tourism subsectors including transportation, shopping, and accommodation, underscores a robust linkage effect. Furthermore, the allocation of digital investments heavily favors tourism transportation and comprehensive services, reflecting a strategic focus that impacts both the economic output and environmental footprint of the tourism industry.

Third, the marked increase in digital investment and burgeoning demand within the tourism sector have significantly driven the growth of China’s DTV and DTE from 2002 to 2017. The refinement of digital application structures since 2007 has emerged as a critical determinant in the uptrend of both DTV and DTE, establishing it as a foremost driver of DEI growth. In contrast, the configurations of digital types and products have moderated the expansion of DTV and DTE. A decrease in carbon emission intensity across tourism subsectors played a crucial role in curbing the increase in DTE, indicating that strategic modifications in digital type and product structures could significantly enhance the positive economic impact of digitalization on the tourism industry. Moreover, optimizing digital application structures while minimizing emission intensity within tourism subsectors could substantially mitigate the environmental impacts associated with digitalization.

Finally, DTV and DTE fell by 61% and 63% between 2017 and 2020, respectively, while the tourism TDI clearly showed a downturn. These sharp drops stemmed from the pandemic’s collapse in demand and strict travel restrictions, which temporarily broke the link between digitalization and tourism—just as Gössling (2023) described COVID-19 impacts as “noise” rather than lasting change [50]. In 2022–2023, China’s tourism spending and carbon emissions quickly returned to their pre-pandemic paths, confirming that these effects were only temporary.

4.2. Policy Implications

To drive sustainable growth in China’s tourism sector, policy should focus on three interlinked priorities.

First, sectoral integration demands a cohesive digital-tourism strategy that raises the level of digitalization across catering and entertainment, while embedding digital-carbon linkage indicators (e.g., DEI, DTE) into national tourism evaluations to spur local governments toward structural reforms.

Second, transportation optimization calls for subsidies to accelerate electric-vehicle fleet adoption and digital dispatch systems, alongside measures to improve energy efficiency and shift to cleaner fuels; piloting “smart tourism zones” with integrated digital–energy platforms in high emission regions such as Yunnan and Hainan will demonstrate how coordinated infrastructure can sharply reduce transport-related emissions.

Third, investment rebalancing requires the tourism industry to calibrate digital-product deployment to avoid overinvestment and ensure efficient use, complemented by targeted tax incentives for green digital solutions. By aligning these measures—sectoral integration, transport decarbonization, and investment optimization—policymakers can harness digitalization to decouple tourism growth from environmental impact.

4.3. Limitations and Future Research

This study presents an integrated supply-driven analytical framework designed to examine the decoupling dynamics between digitization, economic growth, and carbon emissions within the tourism sector. Although this study offers novel empirical evidence on the decoupling effects of digitalization in China’s tourism sector, it is subject to several constraints. First, our analysis relies on national-level data spanning 2002–2020, thereby omitting the post-2020 surge in digital technologies and electric-vehicle adoption that may have reshaped DTV and DTE dynamics. Second, by focusing exclusively on labor and capital within a static input–output framework, we do not account for other critical production factors—such as energy, land, and water—that interact with digital infrastructures and environmental outcomes. Third, the exclusive application to tourism may limit the generalizability of our findings across sectors and economies at different stages of development.

To address the limitations, future research should advance the decoupling framework along four complementary dimensions: (1) spatial scaling, by employing multi-regional input-output models to compare digital-environment coupling patterns both within Chinese provincial systems and between developed versus emerging economies; (2) factor expansion, by incorporating additional inputs—energy, land, water and data—and broadening evaluation metrics from carbon emissions to economic outputs (e.g., GDP, employment) and environmental indicators (e.g., energy consumption, water and land use); (3) sectoral transferability, by adapting the approach beyond tourism to other service industries such as healthcare and education; and (4) methodological enhancement, by embedding Bass diffusion curves [51] within dynamic input–output analyses to simulate post-2020 digitalization and its environmental impacts. Together, these integrated extensions will equip policymakers and researchers with a robust, versatile tool for uncovering how digital transformation can sustainably decouple economic growth from environmental pressures.

Moreover, given the increasing role of AI and automation, future research should explore how emerging AI and automation technologies can further enhance tourism sustainability through three intertwined pathways: (1) predictive optimization, where machine-learning demand forecasts and dynamic pricing algorithms smooth visitor flows and prevent overcrowding—thereby reducing peak-season emissions; (2) autonomous circular systems, in which blockchain-enabled resource-tracking and IoT-integrated, AI-driven energy-management platforms (e.g., smart HVAC and lighting) enable hotels and attractions to minimize waste and optimize resource use in real time; and (3) behavioral nudging, whereby generative AI–powered itinerary planners and personalized recommendation engines guide tourists toward low-carbon activities. By embedding these innovations within dynamic input–output and system-dynamics models, and linking macro-level decomposition with micro-level case studies or experimental designs, scholars can quantify the causal mechanisms through which AI and automation amplify decoupling.