Impact of COVID-19 on the travel and tourism industry

Abstract

Our paper is among the first to measure the potential effects of the COVID-19 pandemic on the tourism industry. Using panel structural vector auto-regression (PSVAR) (Pedroni, 2013) on data from 1995 to 2019 in 185 countries and system dynamic modeling (real-time data parameters connected to COVID-19), we estimate the impact of the pandemic crisis on the tourism industry worldwide. Past pandemic crises operated mostly through idiosyncratic shocks' channels, exposing domestic tourism sectors to large adverse shocks. Once domestic shocks perished (zero infection cases), inbound arrivals revived immediately. The COVID-19 pandemic, however, is different; and recovery of the tourism industry worldwide will take more time than the average expected recovery period of 10 months. Private and public policy support must be coordinated to assure capacity building and operational sustainability of the travel tourism sector during 2020–2021. COVID-19 proves that pandemic outbreaks have a much larger destructive impact on the travel and tourism industry than previous studies indicate. Tourism managers must carefully assess the effects of epidemics on business and develop new risk management methods to deal with the crisis. Furthermore, during 2020–2021, private and public policy support must be coordinated to sustain pre-COVID-19 operational levels of the tourism and travel sector.

1. Introduction

From the start of the COVID-19 crisis in China, the impact of the pandemic on the travel tourism industry was significantly underestimated. Even now, policymakers and tourism practitioners do not have a full understanding of the scenarios and effects of the crisis, which will have an unprecedented impact on the tourism industry. Empirical studies on the impact of pandemic outbreaks on the tourism industry are widely missing in the literature. Our paper is among the first to measure the potential impact of the COVID-19 pandemic in the short and long term, both worldwide and on a geographical level. The study attempts to explore what is expected to be a negative impact on the world and the geographical travel and tourism industry. It also plans to investigate the nature of the impact. Understanding the level of potential impact and the global channels of transmission  will help us predict the extent of current and future epidemic effects on the travel and tourism industry. It will help policymakers and practitioners design policies aimed at capacity building and operational sustainability of the travel tourism sector during 2020–2021 as a policy response to the COVID-19 crisis. Health care quality innovation will play an important role in fighting this pandemic crisis (Zsifkovits et al., 2016).

The latest research report of the world travel and tourism council (WTTC) lists up to 75 million workers at immediate job risk as a result of COVID-19. Research reveals a potential Travel Tourism GDP loss in 2020 of up to US$ 2.1 trillion. WTTC also estimates the daily loss of a shocking one million jobs in the travel tourism sector for the widespread impact of the coronavirus pandemic. Pine and McKercher (2004) researched the impact of the SARS epidemic in 2002 on China's Guandong Province. Their study results show that the impact was negative, substantial, and significant. Mao et al. (2010) studied recovery patterns in Taiwan after the SARS outbreak using a catastrophe cusp model (for details see Sewell et al., 1977; Woodcock and Davis, 1978; Zeeman, 1976) and discovered two factors necessary for recovery from catastrophe to a normal state. The post-recovery period depends on the level of hysteresis and institutional efficiency in facing critical events. Kuo et al. (2008) found that epidemic crises affect tourism demand differently. Results of their study on SARS (2001–2004) and the widespread avian flu (2002–2006) show that SARS had an important impact on tourism demand in China, Hong Kong, Singapore, and Taiwan. The spread of avian flu, however, did not have an important negative impact on the tourism demand in Asia despite the high fatality rate. Preceding studies on pandemic impacts must be extended, however, to account for COVID-19′s new patterns and characteristics. Although in the previous widespread cases (SARS, 2002; H1N1, 2009) inbound tourist arrivals recovered almost immediately after the pandemic alerts were lifted, this will not be the case for COVID-19, and we must take that into account the new empirical models on pandemic impacts. SARS affected Asian stock market integration (Chen et al., 2018) and increased hygienic measures in a limited fashion, but the COVID-19 pandemic seems to demand stricter measures (Kostoff, 2011). Information and communication technologies (ICTs) will play a crucial role in fighting COVID-19 (Gaspar et al., 2019).

To test the impact of COVID-19 on the travel tourism industry worldwide, we set up a dynamic model using an annual data set from 1995 to 2019. The model included 185 countries over 16 different regions and world in total. From the estimated PSVAR models, we take parameters on the empirical link between TCGD, TCEMP, SPEND, GOV, INV, and IPANDEMIC shock. Estimated parameters reflect the empirical link of past pandemic episodes from 1980 to 2019 but say little about the empirical link of the above variables with COVID-19. Demographic patterns in Europe and the rest of the world make population more vulnerable to future epidemic outbreaks (Skirbekk et al., 2015).

To summarize our study results, the impact of COVID-19 on the travel tourism industry will be incomparable to the consequence of the previous pandemic episodes. Depending on the dynamics of future pandemics (from April 2020), the best-case scenario (scenario 1) shows that the travel tourism industry worldwide will drop on average from −2.93 percentage points to −7.82 in the total GDP contribution. Jobs in the travel tourism industry will decrease by −2.44 percentage points to −6.55. The estimated lost inbound tourist spending ranges from −25.0 percentage points to −35.0. Total capital investments that fall due to pandemics varies from −25.0 percentage points to −31.0. The impact is different across regions and in scenarios 1–3, which we further explain in the study.

The paper is structured as follows. After the introduction, Section 2 presents a summary of epidemic outbreaks worldwide since 1980. Section 3 describes the data sample (countries and regions in the sample) and the period used in the modeling process. Section 4 explains the methodological framework of PSVAR with a summary of the results. In Section 5, we develop a system dynamics model that includes the empirical relationship obtained from PSVAR, extended for the new parameters (pattern and dynamic) that resulted from the COVID-19 crisis. We discuss the empirical study results in the Section 6. Section 7 offers concluding remarks on the empirical study results and directions for further research.

2. Epidemic outbreak episodes and tourism worldwide trends since 1980

Several distinct factors determine the impact of an epidemic outbreak on tourism demand. Geographical distance to ground zero (infection epicenter) and infectious power are two of the most distinct. Other modern determinants are media attention (Internet revolution) and associated hysteria. Present worldwide socioeconomic conditions and terrorism, together with conditions of world conflict, impact tourism demand. Oil prices and environmental conditions also exhibit a substantial effect on the number of international tourists travelers. Furthermore, episodes of epidemic outbreaks have coincided with economic turmoil, both nationally and internationally, in the last 53 years (see Table 1 ). Because of tourism's seasonality character and vulnerability to exogenous factors, measuring the impact of an individual factor is a complicated task. This study aims to measure the impact of virus outbreaks on the tourism demand globally since 1980 as a prerequisite to measuring the COVID-19 impact. Several virus outbreaks globally affected worldwide tourism trends and the world economy after World War II (WWII). Table 1 shows the infections and deaths of significant virus outbreaks in the last 53 years.

Table 1.

Infections and deaths of significant virus outbreaks in the last 53 years.

Outbreaks	Infections	Deaths
Marberg (1967)	466	373
Ebola* (1976)	33,577	13,562
Hendra (1994)	7	4
H5N1 bird flu (1997)	861	455
Nipah (1998)	513	398
SARS (2002)	8096	774
H1N1⁎⁎ (2009)	762,630,000	284,500
MERS⁎⁎⁎ (2012)	2494	858
H7N9 bird flu (2013)	1568	616
COVID-19 ⁎⁎⁎⁎ (2020)	1,930,979	120,074

Notes:.

^⁎as of January 31, 2020;.

^⁎⁎between 2009 and 2010;.

^⁎⁎⁎as of November 2019;.

^⁎⁎⁎⁎as of April 2020.

Source: Science Alert (2020).

In the past 50 years, the world has experienced several virus outbreaks with different levels of infections and mortality rates. Fig. 1 plots the virus outbreak episodes to the total number of tourist arrivals by world regions from 1950 to 2018. As expected, the impact varies by world regions depending on the source and distance to the virus outbreak source (ground zero countries).

Fig. 1.

International tourist arrivals by world regions 1950–2018.

Source: Authors' calculation based on data from the World Bank database (2020).

Fig. 1 shows the that different epidemic outbreaks produce different worldwide impacts. Virus epidemic diseases, like SARS (2002) and H1N1(2009), have a large and significant impact on worldwide tourism trends and economic opportunity costs. Epidemic outbreaks with less infectious power (R naught - R ₀ <1) have a lower impact on tourism trends and associated economic losses. Fig. 1 illustrates that the drop in tourist arrivals due to epidemic outbreaks varies across world regions. Table 2 displays registered (direct) drops in the number of tourist arrivals by world regions according to data from the United Nations World Tourism Organization (UNWTO)—a United Nations specialized agency—database.

Table 2.

Tourist arrivals and spending in a time of epidemic outbreaks by world regions, 1980–2019.

Regions	Crisis date	International tourist arrivals (in Mn.)	Change over the previous year (%)	Lost arrivals (in M)	Lost spending (US$ bn)
Africa	Hendra (1994)	18	2	–	–
	H5N1 bird flu (1997)	21	5	–	–
	Nipah (1998)	24	10	–	–
	SARS (2002)	29	5	–	–
	H1N1⁎⁎ (2009)	46	4	–	−2
	MERS⁎⁎⁎ (2012)	52	5	–	–
	H7N9 bird flu (2013)	55	4	–	–
Americas	Hendra (1994)	105	3	–	–
	H5N1 bird flu (1997)	116	1	–	–
	Nipah (1998)	119	3	–	–
	SARS (2002)	106	−2	−3	–
	H1N1⁎⁎ (2009)	141	−5	−7	−21
	MERS⁎⁎⁎ (2012)	163	4	–	–
	H7N9 bird flu (2013)	170	5	–	–
Asia and the Pacific	Hendra (1994)	80	11	–	–
	H5N1 bird flu (1997)	90	−1	−1	−2
	Nipah (1998)	89	0	–	–
	SARS (2002)	113	−9	−12	−2
	H1N1⁎⁎ (2009)	184	−1	−3	−6
	MERS⁎⁎⁎ (2012)	238	7	–	–
	H7N9 bird flu (2013)	254	7	–	–
Europe	Hendra (1994)	297	4	–	–
	H5N1 bird flu (1997)	350	7	–	–
	Nipah (1998)	360	3	–	–
	SARS (2002)	416	2	–	–
	H1N1⁎⁎ (2009)	473	−5	−26	−61
	MERS⁎⁎⁎ (2012)	539	4	–	–
	H7N9 bird flu (2013)	567	5	–	–
Middle East	Hendra (1994)	11	9	–	–
	H5N1 bird flu (1997)	16	7	–	–
	Nipah (1998)	17	6	–	–
	SARS (2002)	27	4	–	–
	H1N1⁎⁎ (2009)	49	−5	−3	–
	MERS⁎⁎⁎ (2012)	52	3	–	–
	H7N9 bird flu (2013)	51	−2	−1	−1

Source: Authors' calculations on the data the from UNWTO database.

Table 2 shows that the total lost tourist arrivals worldwide from 1980 to 2019 amounted to 57 million (M) during the epidemic outbreaks. Lost tourism spending worldwide in times of epidemic outbreaks during this same period reached 95 US$ billion (bn). In relative terms, total lost tourism spending in a time of epidemic crisis was 0.23% of the world GDP (to the average world GDP value from 1980 to 2018). Epidemic outbreaks vary significantly between the type of disease outbreak and across world regions. Africa did not experience a significant impact on the tourism demand during the epidemic crisis; total losses were 2 bn US$ in tourism spent from 1980 to 2019.

The SARS epidemic crisis of 2002 and H1N1 (2009) caused a striking drop in tourist arrivals by −10 million in the Americas region; in tourism spending, the loss was −21 bn US$ in the that region. The Asiatic and Pacific regions experienced a significant drop in tourist arrivals during the bird flu epidemic (1997), SARS (2002), and H1N1(2009). Lost arrivals during the bird flu crisis in that region was −1 million, and lost spending amounted to −2 bn US$ . The decline in the tourist arrivals in the region at the time of the SARS (2002) outbreaks was −12 million with a related −2 bn US$ in lost revenue. During the H1N1 (2009) crisis, the region experienced −3 million tourist arrivals decline and −6 bn US$ lost tourism spending. The European region was not significantly hit by most of the outbreaks and epidemics from 1980 to 2019 (for the distance to the virus originating region). However, in the H1N1 epidemics (2009), there was a decline of −26 million tourist arrivals and a −61 bn US$ total tourism spending loss (amounting to 0.5% of the Europe GDP at the time). In H1N1 epidemic episode, however, Europe was affected strikingly with a total decline in tourist arrivals and spending loss that surpassed the economic impact for all other world regions over the 1980–2019 period.

The Middle East region suffered travel disruptions during the H1N1 (2009) epidemic crisis, causing a −3 million decline in tourist arrivals. Tourist spending, however, did not register a decline due to a rise in the receipts per arrival. The bird flu (2013) crisis faced a −2 million decline in tourist arrivals and −1 bn US$ in international visitor spending.

We associate epidemic outbreaks with significant opportunity costs in tourism demand. Although epidemic outbreaks significantly shape and influence the tourism industry, multiple causality issues (e.g., outbreaks followed by environmental and security issues or political and economic crises) arise when it comes to measuring the opportunity costs of epidemic outbreaks. To address this issue, we use structural vector auto-regression (SVAR) models to estimate the direct and opportunity costs of COVID-19 on the tourism industry (and economy) worldwide.

3. Data and facts on pandemics' impact on travel and tourism

The travel and tourism industry in the time of services-led growth trends has become increasingly important worldwide since 1990. From 1995, the travel and tourism industry's direct contribution to the world GDP increased from 9.9% in 1995 to 10.3% in 2019. A significant impact of the travel and tourism industry is also visible on employment levels. The total contribution to world employment in 2019 was 10.4%. The SARS epidemic in 2003 and the financial crisis of 2008 had a significant negative impact on world and regional travel and tourism industries.

We use annual data for 185 countries grouped in 16 world regions: Africa, the Americas, Asia Pacific, Caribbean, Central Asia, Europe, Latin America, Middle East, Northeast Asia, North Africa, Northern America, Oceania, other Europe, South Asia, Southeast Asia, and sub-Saharan Africa. All data are in real prices (CPI US$ 2000=100 index), adjusted for the impact of inflation. We utilize two main databases: World travel and tourism council (WTTC) data gateway (wttc.org/datagateway) and UNWTO (unwto.org/data). All data are in annual frequencies from 1995 to 2019. Data series (variables) we use for modeling:

• •TCGDP = total (indirect and induced impact) of travel and tourism contribution to the national/regional GDP (in real US$ bn, WTTC 2020)
• •TCEMP = total (indirect and induced impact) travel and tourism contribution to national/regional employment (in 000 of jobs, WTTC 2020)
• •SPEND = total spending in the domestic economy by foreign visitors (in real US$ bn, WTTC 2020),
• •ARRIVALS = total tourist arrivals (in 000, UNWTO 2020)
• •GOV = individual government expenditures on travel and tourism (in real US$ bn, WTTC 2020)
• •INV = Investment - capital investment both private and public (in real US$ bn, WTTC 2020)
• •PANDEMIC = dichotomous (dummy) variable, PANDEMIC = 1 when no pandemic outbreaks exist and PANDEMIC = 0 when pandemic outbreaks are present
• •IPANDEMIC = measures the impact of pandemic outbreaks SARS (2002), H1N1 (2009), MERS (2012), and H7N9 bird flu (2013)

To correctly capture the pandemic impact, we use the interaction variable (model) with ARRIVALS as a continuous variable (000 of jobs) and PANDEMIC as a dichotomous variable (dummy variable 0,1).

4. Measuring pandemics' impact on the tourism industry using PSVAR

Our goal is to analyze the impact of the new coronavirus (COVID-19) outbreak on world tourism dynamics. To our knowledge, this study is the first attempt to measure the impact of the current COVID-19 outbreak on tourism worldwide. To estimate the impact of COVID-19′s ongoing outbreak, we model the COVID-19 shocks using time series data for past coronavirus outbreaks; for SARS (2002), H1N1 (2009), and Ebola (2004); and for the past epidemic outbreaks of Hendra (1994), H5N1 bird flu (1997), Nipah (1998), MERS (2012), and H7N9 bird flu (2013). However, COVID-19 is a new type of virus, and it brings much uncertainty connected to the speed of spread, infectious power, mortality rate, and future dynamics of the virus. Therefore, we use the current state of knowledge on the COVID-19 reproduction number (R0) (Kucharski et al., 2020) to calibrate our models for the SARS (2002) and H1N1(2009) variance. To study the impact of COVID-19 on tourism worldwide, we use the heterogeneous PSVAR model as developed by Pedroni (2013). Using the PSVAR model—Ewiews code provided by Luvsannyam (2018) and Góes (2016)—we estimated how the COVID-19 shocks on tourism (arrivals, spending) propagated across world regions. We have a strong heterogeneous sample, so using PSVAR enabled us to estimate the impact of the COVID-19 shock on tourism, depending on the following factors: region-specific socioeconomic conditions, vulnerability to external shocks, health system stability, environmental conditions, tourism sector stability, and competitiveness.

Following Pedroni (2013), we decomposed the impact of the structural shock (COVID-19) into common shock (effects of COVID-19 outbreak originating in any other region in the sample and propagating to region-specific tourism industry) and idiosyncratic shock.

Idiosyncratic shocks show the impact of COVID-19 originating in a member-specific region on the tourism industry in the same region. As in Biljanovska et al. (2017) and Pedroni (2013), we refer to the measured COVID-19 effects on the tourism industry as a common component or spillover (common shock), and to an idiosyncratic component as country-specific (idiosyncratic shock).

We use the following estimation model for an unbalanced panel using data from 1995 to 2019 in the PSVAR bivariate form (for details see Góes (2016); Pedroni (2013); Biljanovska et al. (2017):

$$B_{i}y{*}_{i,t}=A_i\left(L\right)y{*}_{i,t-1}+{\epsilon }_{i,t}$$ (1)

where y*_i,t is an s n-dimensional vector of demeaned stacked endogenous variables, and

$$A_i\left(L\right)\equiv \left(\sum _{j=0}^{J_i}{A^i}_j{L}^j\right),$$ (2)

which equals polynomial of lagged coefficients with country-specific lag lengths J_i, coefficient matrix Aⁱ_j, vector of stacked residuals e_i,t and B_i contemporaneous coefficient matrix (Góes, 2016).

In this model, we allow for full heterogeneity and robust inference decomposing (using time effects) impulse response function to idiosyncratic shocks and common shocks (region-specific response). To estimate the impact of pandemic outbreaks, we use TCGDP, TCEMP, SPEND, ARRIVALS, GOV, and INV as endogenous variables, and PANDEMIC (dummy), IPANDEMIC (interaction variable), defined in Section 3. Structural shock (white noise vector) takes the form

$${\xi }_{it}={\left({\overline{\epsilon }}_t^{\prime },{\tilde{\epsilon }}_{it}^{\prime }\right)}^{\prime },$$ (3)

with ${\overline{\epsilon }}_t^{\prime }$ = vector of common white noise and ${\tilde{\epsilon }}_{it}$= vector of idiosyncratic white noise shocks.

Composite white noise errors equal

$${\epsilon }_{it}={\Lambda }_i{\overline{\epsilon }}_t+{\tilde{\epsilon }}_{it}$$ (4)

and Λ_i = M x M diagonal matrix with loading coefficients Λ_i, m = 1,…, M (Pedroni, 2013).

Composite structural white noise shock (vector ε_it) takes the form

$${\epsilon }_{i,t}^{IPANDEMIC}={\lambda }_i^{IPANDEMIC}{\overline{\epsilon }}_{.,t}^{IPANDEMIC}+{\tilde{\epsilon }}_{i,t}^{IPANDEMIC}$$ (5)

$${\epsilon }_{i,t}^y={\lambda }_i^y{\overline{\epsilon }}_{.,t}^y+{\tilde{\epsilon }}_{i,t}^y,$$ (6)

with ε_it = composite white noise shock, ε_.,t = common shock, and ${\tilde{\epsilon }}_{i,t}$ = country-specific idiosyncratic shock (Biljanovska et al., 2017).

We use long-run identifying restrictions (Blanchard-Quah, 1989) to obtain structural shock estimates for composite and common shocks. Impulse responses of common and idiosyncratic shocks (rescaled) show impulse responses to unit shocks. We intend observing an impulse response to a unit shock (−100,000 tourist arrival drop) in IPANDEMIC when pandemic outbreaks are present.

Fig. 2 presents the median composite response of the total (indirect and induced impact) travel and tourism contribution to the national/regional GDP (TCGDP) in a pandemic outbreak (unit shock) for 17 regions (including world) of the sample. A large negative impact of a pandemic outbreak on the (TCGDP) GDP created by the travel and tourism industry can be observed. The initial response of the TCGDP is negative for all regions and the world in total (panel a).

Fig. 2.

Impulse (composite) response functions’ estimate of TCGDP to IPANDEMIC

Notes: Median impulse response averaging with general to specific lag length criteria selection and 95% confidence intervals.

Source: Authors' calculation

When a pandemic outbreak occurs, for a decline of −100,000 tourist arrivals (unit shock for pandemic), the total GDP created by the travel and tourism industry drops by −0.46 (bn) US$. The negative impact is largest in the first year with −0.46 bn US$ of the GDP lost because of the pandemic outbreak. The large initial negative impact is offset in the next year, declining from −0.46 bn to just −0.01 bn US$ in the second year. Thus, policymakers and travel and tourism practitioners can expect significant losses (in terms of the GDP created by tourism) in the first year of a pandemic outbreak. If the impact is larger than a unit shock, for example, a 10-unit increase in IPANDEMIC (equal to 1 million tourist arrivals lost) will cause −4.6 bn US$ decline of the real GDP contributed by the travel and tourism industry. With a 20-unit increase in IPANDEMIC (2 million tourist arrivals lost), we can expect the associated real GDP will drop by −9.2 bn US$. Policymakers and travel and tourism practitioners must know the deep potential impact of pandemic outbreaks on the tourism and travel industry and the national economy. Although the negative impact of the pandemic is limited to the medium-term effect (see the Fig. 2) on the tourism and travel industry and national economy, the impact is too large and sizable to ignore. It demands an appropriate managerial and economic policy response, an immediate and direct comeback to avoid potentially catastrophic consequences. The response must be straightforward and quick, since the time of response and the extent or limit of the measures will dictate the actual level (negative peak) of the pandemic shock. Without the appropriate managerial and economic policy response, swift and directed, the pandemic shock consequences are large enough to push any economy (even the most advanced) into a deep recession. The potential impact of a pandemic is great enough to have a tsunami effect on the total travel and tourism industry, leading to large-scale bankruptcies. We can observe the double-dip pandemic shock effect since after the initial shock TCGDP in the second-year decline just by −0.01 bn US$ for −100,000 tourist arrivals lost. However, in the third year, the drop in the TCGDP amounts to −0.06 bn US$ per −100,000 tourist arrivals lost. The pandemic shock effect dies out in the fourth year after the initial shock (medium-term effect). The cumulative negative impact of a unit pandemic shock (a drop of −100,000 arrivals) to the travel and tourism industry and the national output is −0.53 bn US$ (decline) of the revenue that the industry contributed the GDP. Pandemic crises because of their total potential economic impact on the travel and tourism industry and the national economy should be classified as a super cycle event in economic literature. For example, a pandemic event resulting in −500 million tourist arrivals' drop would cause a −2.809 bn US$ drop (3.3% of the world GDP 2018) in the GDP contributed by the travel and tourism industry worldwide.

Fig. 2 illustrates the estimate (common shock) response to the impact of the pandemic effect that originates in any other region in the sample (pandemic cluster), spilling over to the region of interest (see panel b in Fig. 2). The impulse response to regional pandemic effect (IPANDEMIC unit shock) is less pronounced with the composite shock impact. Past pandemic outbreaks over regions in the sample show an important but limited impact on a specific region of the travel and tourism industry and national output. Pandemic outbreaks show limited (spillover) impact when compared to the composite effects of the IPANDEMIC shock.

A unit IPANDEMIC shock (−100,000 tourist arrivals) harms the GDP originated by the travel and tourism industry of −0.10 bn US$. This means that a more extensive pandemic shock (a drop of −1 million tourist arrivals) would cause a −1 bn US$ decrease in the GDP contributed by the industry. Common shocks' impact on a particular region is less pronounced regarding the composite shock (worldwide impact), and it can have significant but limited negative effects. Remember, the composite shock impact of a unit IPANDEMIC shock is −0.46 bn US$ compared to −0.10 bn US$ for the common shock, which means that not all regions will be in danger of a super cycle event triggered by the pandemic crisis. The negative impact will not be at an "economic extinction" level but will still be important to observe. It can be observed that the median impact of the pandemic outbreak reaches through during the first year the crisis appears. The impact of the crisis is statistically significant (inside 95% confidence intervals – dashed lines). It offsets the shock effects in the second year (dropping to zero) but with a double-dip negative impact as for composite shock. The negative effect reappears in the third year with a −0.03 bn US$ impact on the GDP generated in the travel and tourism industry. We can conclude that some regions show more resilience to the negative impacts of the pandemic shock. Such evidence demands more research on the factors that determine the regional vulnerability to a pandemic crisis (like the phase of the financial/business cycle, government response, micro and macroeconomic stability, institutional efficiency, public sector efficiency, and health system stability). While the pandemic crisis is potentially devastating on a worldwide level, some regions are less vulnerable than others, and the regional economic policy is an important tool in fighting pandemic shocks.

Idiosyncratic impulse responses show how vulnerable regions are to the pandemic shock originating within the same region (idiosyncratic shocks). Panel (c) in Fig. 2 explains the effects within the region pandemic event. As expected, pandemic events originating within a particular region have devastating economic impacts on the same region. The median impact reaches through during the first year, attaining −0.45 bn US$ per unit IPANDEMIC shock. Pandemic outbreak within a region has an immediate negative effect on the GDP created by the travel and tourism industry and national output. A unit shock (−100,000 arrivals) results in the decline of the GDP (GDP drop) by −0.45 bn US$. The reaction to the shock is swift and profound. During the second year, the shock effects die out, reaching −0.01 bn US$, but the double-dip effect is present (the same as for composite and common shock). After "economic reanimation" in the second year, the negative effects of the shock intensify, reaching −0.05 bn US$ of TCGDP. The total effects of the shock die out four years after the shock, with the impulse response converging to zero. The estimated impulse response of the TCGDP to a unit shock is statistically significant over the whole period. We observed that idiosyncratic shocks of epidemic outbreaks (region as an epidemic cluster) were more sizable in relation to the common shocks. Idiosyncratic shocks are more important than common shocks, with composite shocks driving the dynamics of the idiosyncratic shocks. Pandemic outbreaks starting within a region impact economic activity connected to travel and tourism. National pandemic outbreaks within a region have a larger negative impact on the region's economic activity than the spillover shocks coming from pandemic outbreaks in other regions. The consequences of domestic pandemic shocks are significantly larger with more repercussions than spillover shocks in other regions of the world. Even in a globalized world economy, pandemic outbreaks in one region will have limited impacts on the tourism industry in other regions. However, a pandemic crisis started in a particular region will have a large and significant negative effect on the same region's domestic economy.

To conclude, pandemic shocks have a deeply negative impact on the tourism industry and economy. National pandemic outbreaks (idiosyncratic shocks) are significantly larger and more damaging to outside pandemic outbreaks (common shocks). Spillover negative effects of a pandemic outbreak have a limited impact on the tourism industry and economy as long as the pandemic outbreak started elsewhere does not become a "domestic" pandemic cluster. The difference in the extent of the common versus idiosyncratic shocks consequences comes from the differences in the domestic (region's) tourist industry's diversity and stability, the nation's/region's economic resilience, and the structure and level of technological and institutional development.

Fig. 3 examines a much different pandemic impact: the shocks on the employment dynamics (TCEMP) and the reaction of employment (TCEMP) to a unit shock (IPANDEMIC) over ten years.

Fig. 3.

Impulse (composite) response functions’ estimate of TCEMP to IPANDEMIC

Notes: Median impulse response averaging with general to specific lag length criteria selection and 95% confidence intervals.

Source: Authors' calculation

The median response of employment to the pandemic shock reaches through during the first year (negative effect). The effects of the composite shocks (panel a) are negative and statistically significant (within the 95% confidence intervals). Composite shocks (sum of the common and idiosyncratic) to employment (direct + indirect employment created by the travel and tourism industry) show a pattern similar to the GDP and the tourism industry. The negative impact of a pandemic shock is significant and sizable, with an immediate decline in the employment level during the first year. A one-unit shock (decrease of −100 tourist arrivals) for a pandemic outbreak results in −24 lost jobs contributed by the travel and tourism industry. Thus, the pandemic shock has a significant negative impact not only on the employment level in the travel and tourism (direct) but also on associated sectors in the economy (indirect impact). For a −100,000 drop in tourist arrivals, direct and indirect employment created by the travel and tourism industry declined by −24,000 jobs. A more pronounced shock, a 10-unit increase in IPANDEMIC (one million fewer tourist arrivals) is associated with a decline of - 240,000 jobs in the travel and tourism industry and associated sectors. As expected, epidemic outbreaks have potentially catastrophic negative impacts on the industry as well as the whole economy. A large pandemic outbreak (similar to COVID-19) could cause −10 million tourist arrivals, triggering massive layoffs in the travel and tourism industry and associated sectors by −2.400,000 jobs. A massive −100 million decline in tourist arrivals worldwide (7.14% of the total world overnight visitors in 2018) could start a chain reaction leading to −24 million lost in travel and tourism worldwide (7.54% of the total travel and tourism contribution to the employment in 2018).

Negative effects slow down during the second year, recovering from the initial shock and dampening to zero. The total effect of the pandemic shock is highly persistent, with pandemic shock effects converging to zero after five years. Still, such a large pandemic shock does not show optimistic signs that the travel and tourism industry could return to pre-pandemic employment levels rapidly. We expect the industry to recover slowly during the initial five years, depending on the overall pandemic shock impact to the economy.

Panel (b) in Fig. 3 shows the spillover effects of a pandemic shock starting in another region and spilling over to a specific (domestic) region. Unlike the pandemic impact on the GDP spilling over to other world regions, labor markets associated with the travel and tourism industry are highly globalized and volatile. Common spillover shock effect is large and statistically significant (within 95% confidence limits), reaching through in the first year (immediate negative impact) of the pandemic episode. For a −100,000 tourist arrival decline, it can be expected that −18,000 jobs (direct + indirect effects) will be lost. Common shock dynamics follow composite shock dynamics, reaching through in the first year and slowing down with a modest recovery in the second year. The pandemic effect on employment contributed by the travel and tourism sector worldwide remains highly persistent over five years. Large pandemic shocks, like −100 million lost tourist arrivals, would result in −18 million laysoff worldwide (5.67% of the total world contribution of the travel and tourism sector to employment in 2018). Pandemic shocks have a sizeable negative effect on the highly globalized tourism labor market, as we can see from Fig. 3. Pandemic shock originating in one region and having immediate negative effects on the domestic labor market is rapidly and extensively spilling to other world regions' labor markets (travel and tourism). A catastrophic pandemic crisis leading to 1 bn lost tourist arrivals would devastate the travel and tourism industries worldwide, causing −180 million lost jobs (56.7% of the world's total travel and tourism contribution to employment). Half of the travel and tourism industry and associated sectors worldwide would face economic collapse. Common shocks resulting from the pandemic crisis show significant spillover, transferring negative effects of the crisis from the domestic tourism labor market to the foreign. Global labor markets in the tourism industry are more vulnerable and volatile to the spillover effects of the pandemic crisis compared to the spillover effects on national economies.

A pandemic crisis starting (domestic) idiosyncratic shock (panel c in Fig. 3) has a negative and statistically significant impact on employment. The same pattern as in the composite and common shock dynamics is also visible in the idiosyncratic shock (domestic) in the region. Pandemic outbreaks starting within a region have an immediate negative impact on the travel and tourism sector at the employment level. The median impact is large and negative, reaching through in the first year and swiftly dying out in the second year. In the third year, the negative effect on employment reappears, converging to zero in the next two years. The impact of the pandemic crisis on employment in the domestic market is persistent and shows a medium-term pattern. The case is similar in the common shock; a domestic pandemic cluster (unit shock in IPANDEMIC) will bring down employment in the travel and tourism industry and associated sectors. For −100 lost tourist arrivals, −14 jobs will be lost in the domestic (regional) tourism industry. Although the negative impact does not appear significant at the start (same as an illusion with the linear and exponential growth), its cumulative effects are staggering. A total of −140,000 job positions will disappear in the region due to a regional pandemic crisis, causing - 1 million tourist arrivals. Over a decline of −100 million lost tourist arrivals to a region, the total number of job positions strayed will hit −14 million. Such a shock will bring a sizable negative effect on the region. For example, in Africa (in 2018) the travel and tourism industry contributed to 24.3 million jobs. Losing −14 million in the case of Africa, with a total of 24.3 million job positions related to tourism, would be devastating (−57.6% of the total jobs in Africa's tourism industry). The shocks starts slowly (from −14 job positions lost) and rapidly grows to −14 million, bringing a catastrophic unemployment crisis to the tourism industry and economy. From the impulse response function, we observe that no region is immune to the pandemic impact on the domestic tourism labor market. However, the final level of the negative impact depends on the importance of the tourism industry in the economy (share in the GDP) and other micro/macroeconomic factors associated with institutional and economic policy efficiency.

When we compare the median and the average (mean) composite, common, and idiosyncratic responses of employment to the pandemic crisis, the level of impact varies across regions (figures not presented here for space constraint). Further study on the determinants of heterogeneous responses across regions regarding employment to the pandemic impact (also at a national level) demands attention to better understand the mechanism behind pandemic shocks on the tourism industry.

Pandemic shock (IPANDEMIC) results in the decline of tourist arrivals transmitted to tourist spending (SPEND). Fig. 4 illustrates the negative impact of the pandemic crisis on foreign visitor tourist spending. First, we look at the composite shock impulse response (panel a), which resembles the pattern we discovered for TCGDP and TCEMP. The negative impact of the shock reaches through during the first year (immediate effect), reversing to normality during the second year. In the third year, we can see a second negative impact slowly converging to zero during the next two years. A unit shock in IPANDEMIC (−100,000 tourist arrivals) results in a −0.24 bn US$ decline in tourism spending. After the peak in the first year, the median impulse response of tourist spending to the pandemic shock declines but continues to persist in the medium term (five years). Unlike the response of employment to the pandemic crisis and similar to the output response, tourist spending response to a unit shock is more pronounced in the case of idiosyncratic shocks.

Fig. 4.

Impulse (composite) response functions’ estimate of SPENDING to IPANDEMIC

Notes: Median impulse response averaging with general to specific lag length criteria selection and 95% confidence intervals.

Source: Authors' calculation

Panel (b) in Fig. 6 shows that the effect of the common shocks to a unit shock (−100,000 tourist arrivals) is −0.07 bn US$ of lost tourist spending. The common shocks impact is less pronounced to the composite, meaning that spillover effects of the pandemic crisis from one region to another are limited. Tourism service markets are highly globalized, but, according to our results, more robust and less volatile to pandemic shocks compared to the tourism labor markets. Pandemic crises that originated in other regions in our sample have a limited impact on foreign tourist spending in the domestic region. The impact, as shown in Fig. 4, is statistically significant at a 95% confidence interval, reaching a peak in the first year and converging to zero in 3 years. Double-dip impact is less pronounced when compared to the impact on output or employment. We can conclude that the spillover impact of a pandemic crisis is displaced from one world region to another but is less profound when the pandemic impact starts within the region. A pandemic outbreak starting within a particular (domestic) region (idiosyncratic shock in panel c) exhibits more striking negative effects on tourism spending than a pandemic impact migrating from outside regions. The impact of a pandemic unit shock measured by impulse response (idiosyncratic shock) shown in Fig. 4 points out that tourist spending declines by −0.22 bn US$ for a decrease of −100,000 tourist arrivals. Compared to the spillover effects (common shocks) of −0.07 bn US$, we can see that domestic pandemic crises (response to idiosyncratic shocks) have three times (−0.22 bn US$) the number of devastating effects on tourist spending. The median impulse response reaches through during the first year after the shock, converging to zero in the second year with a double-dip in the third year. The pandemic impact of the idiosyncratic shock on tourist spending persists in the medium term (five years). Domestic pandemic outbreaks hit the domestic tourism sector and economy hard. For a −1 million of lost tourist arrivals, the domestic travel and tourism industry and economy loses −2.2 bn US$ in tourism spending. Larger crises, such as COVID-19, that could cause −100 million lost tourist arrivals could also cause regions to lose about −220 bn US$ (14% of the world tourism spending in 2018).

Fig. 6.

Estimated potential (COVID-19) impact on employment in the travel and tourism industry worldwide.

Source: Authors’ calculations using their system dynamic model

Impulse response estimates of the composite, common, and idiosyncratic pandemic shocks on tourism created the GDP, employment, and tourism spending present median responses. Average responses vary across regions, meaning regions show variances in the response estimates (the difference between the median and average estimates to shocks). This study does not present figures for the mean response estimates to shocks due to space constraints. Using PSVAR, we also assess median response estimates to pandemic shocks to capital investments: INV = Investment - Capital investment, both private and public (in real US$ bn), and government investments (GOV) = government individual expenditures on travel and tourism (in real US$ bn). Including these PSVAR estimates allows us to access the total impact of the pandemic crisis on the overall travel and tourism industry. Since calculations demand numerous tables and figures, we do not present them here due to publishing constraints, but we do give a summary and discussion in the concluding remarks.

Pandemic outbreaks have potentially devastating effects on the travel and tourism industry worldwide according to the evidence obtained from the PSVAR models in this paper. Policymakers and practitioners in the tourism industry should pay significant attention and adapt future economic and management policies accordingly. Empirical evidence of this study supports this thesis, providing robust evidence to hold the pandemic shocks' theory. PSVAR relies on structural and reduced-form modeling and historical series data to obtain empirical links between pandemic shocks and main indicators in the travel and tourism industry.

However, since COVID-19 is a new pandemic outbreak for which we do not have reliable historical series data, we have to estimate the economic impacts of COVID-19 on the tourism industry worldwide using empirical knowledge gained with PSVAR modeling. The same PSVAR empirical knowledge gained in this study gives us the empirical background we need to develop a business dynamics model to test the economic impact of the current COVID-19 pandemic crisis.

5. Estimating COVID-19 pandemic shock economic impact on the world and the regional travel and tourism industry

To evaluate the impact of COVID-19 on the travel and tourism industry worldwide, we set up a dynamic model (not presented here due to space constraints). From the estimated PSVAR models, we take parameters on the empirical link between TCGD, TCEMP, SPEND, GOV, INV, and IPANDEMIC shock. Estimated parameters reflect the empirical link from past pandemic episodes from 1980 to 2019 but say little about the empirical link of the above variables (COVID-19). To measure the potential impact of COVID-19, we recalibrate estimated parameters to correct for the knowledge we now have on COVID-19. Our dynamic model includes recalibrated PSVAR parameters, R₀ for COVID-19, a proxy for government responses (Hale et al., 2020), country's/region's economic policy responses (RBA Research International), the share of export in the GDP, the travel and tourism sector share in the GDP, the phase of the financial cycle (credit-to-GDP gaps - BIS data), private debt share in the GDP, and tourist arrivals under three scenarios. We set up the number of international tourist arrivals under three different scenarios.

The first scenario is a lockdown as it occurred during March 2020 and continuing in April (scenario one from January 1, 2020 to April 1, 2020). The second scenario projects the continuation of the lockdown from April 1, 2020 to August 1, 2020. The third and worst scenario of the pandemic outbreak projects it staying in the environment until the end of 2020 (scenario three from August 1, 2020 to December 31, /2020). We use the above-explained building blocks to build a dynamic model for estimating the potential impact of COVID-19 on the travel and tourism industry worldwide.

Fig. 5 shows the estimate of the potential COVID-19 impact on the travel and tourism industry at the world and regional levels (regions according to the WTTC) under different scenarios. The first scenario is already in place (actual scenario) since we are well beyond its time duration: January 1, 2020 to April 1, 2020. This is the best-case scenario, assuming the pandemic outbreaks to be under control during April 2020 and a complete worldwide lockdown is revoked.

Fig. 5.

Estimated potential (COVID-19) impact on output in the travel and tourism industry by world regions.

Source: Authors’ calculations using their system dynamic model

The numbers in Fig. 5 present losses in the real GDP (TCGDP) in the travel and tourism industry (in US$ bn 2000 constant prices). Numbers in Fig. 5 stand for the GDP the travel and tourism industry could create but lost due to the outbreak of COVID-19. Under scenario 1, the world will suffer economic costs equal to −2.2 trillion US$ or −4.54% of the world's GDP. The actual scenario (scenario 1) tells us we can expect the world's GDP to fall by −4.54% in 2020. The world will experience a new recession phase amounting to a −4.54% GDP decrease if scenario 1 persists. We can see that pandemic economic shocks are significant and substantial (even under the best actual scenario). The costs are much higher even now under scenario 1 than policymakers and practitioners expected when COVID-19 first appeared in China. Thus, we can call scenario 1 an actual or inevitable economic cost that the world will face because of the COVID-19 outbreak.

Fig. 5 shows how the pandemic economic shock spillover effects vary across world regions. Advanced economic regions like the Americas, Europe, NortheEast Asia, Asia Pacific, and Northern America will experience significant declines in the GDP, from −853 bn US$ in Europe to −895.6 in Northeast Asia. The Americas and the Asia Pacific will face a significant decline in the GDP, with the Americas losing 1.5 trillion US$ and Asia Pacific 1.1 US$ trillion. In absolute numbers, other world regions will face a significantly lower amount in terms of dollar amounts, but in relative terms (their share in the lost GDP), they will face the same recession pressure as advanced regions. The region's potential recession is a consequence of a domestic pandemic shock and limited to a minor spillover effect of the pandemic shock from other regions. The reason lies in the economic lockdown policies that regions adopted to fight COVID-19. However, common shocks or spillover effects for COVID-19 are significantly more substantial when compared to the same effects for SARS (2002), H1N1 (2009), or MERS (2012). According to the actual situation and actual scenario (scenario 1), regions will face recession ranging from a decline in the GDP from −2.09% (Central Asia) to −7.82% (Caribbean).

Scenario 2 describes an undergoing scenario (April 2020). It shows that the adverse macroeconomic effects of the COVID-19 pandemic crisis are much worse than policymakers and analysts expected when the crisis started. With the COVID-19 outbreak and lockdown continuing to August 1, 2020, the world economy will face a decline in output of −9.80%. Again, advanced regions will be affected the most (due to a drop in a large number of tourist arrivals) with the Americas facing −10.93%, Europe −8.33%, and Southeast Asia −9.88%. However, less developed regions like Africa (−10.08%) and North Africa (−10.03) could also experience significant recession. For less-developed regions, the dynamics of tourist arrivals are somewhat different (due to climate and environment), so their economic losses are more significant under the second scenario than under the first. Other regions, on average (−8.69%), will experience a pandemic shock with an output drop ranging from the lowest, - 4% in Central Asia, to the highest, −16.34% in the Caribbean and −9.59% in Oceania.

The third scenario, from August 1, 2020 to December 1, 2020, is the more severe scenario in our model (if lockdown continues to end of this year). The average drop in output under scenario 3 equals −12.72% for all regions in the sample. Regions with a significant share of tourism contribution to output, as in the Caribbean (−23.69%) and Oceania (−14.97%), will face significant idiosyncratic shocks impact. The less-developed regions of Africa (−14.95%) and North Africa (−14.97%) will experience a cascade effect resulting from both idiosyncratic and common shocks disrupting the supply–demand mechanism. Such regions will experience a double impact resulting from the domestic effects of the outbreak on the economic activity that amplified the worsening economic conditions abroad. Total world output if the lockdown continues to the end of 2020 will reach levels (−14.20%) far worse than the 2008 economic crisis and constitute the biggest plunge after WWII. The plunge depends mainly on the lockdown conditions that, if relaxed, will move recession scale more toward scenario 2.

Fig. 6 shows estimates the potential impact of COVID-19 on the labor (and associated) markets in the travel and tourism industry.

Fig. 6 also shows that the connected industries’ labor market will be significant and negative. As expected, the impact will be more substantial in regions more dependent on the travel and tourism industry. The Asia Pacific region under scenario 1 will lose 56.6 million jobs due to COVID-19. A large part of the region's travel and tourism industry, with 186 million jobs, will suffer a significant shock if not counterbalanced by government measures to fight the adverse effects of COVID-19 on the labor markets. Potential negative impacts multiply as lockdowns continue or countries retain strictly controlled border regimes (as announced by France and other EU members, at least to September 2020). Under these conditions, the second scenario applies, with the Asia Pacific regions losing a million jobs in the travel and tourism industry. Under the first scenario, the Americas could loss −19.9 million jobs, an actual scenario if we observe the last data for the United States with more than 15 million unemployment claims by March 21, 2020. In Canada, jobless claims in the same period peaked at 2.13 million with the model predictions fitting the data quite well. Northeast Asia could experience −38.4 million jobs lost in the travel and tourism industry under the first scenario, with South Asia reaching −22.4 million. The impact will be on a “tsunami” level for the Caribbean, losing almost 46% of the total jobs in the industry. Scenario 1 will hit the world by large force with −164.5 million jobs lost in the travel and tourism industry. Although scenario 1 is the actual scenario now (April 2020), the estimated impact is ample, but recovery is still possible since the average recovery time in the travel industry is estimated between 10 and 12 months. However, if the lockdown continues with travel limitations imposed, the second scenario could bring severe challenges to the travel and tourism sector well beyond the sector's resilience threshold.

Under scenario 2, jobless claims worldwide could hit 354.7 million with direct jobless claims in the travel and tourism labor markets reaching −118 million. Regions experiencing the most significant declines in employment (in absolute numbers) will be Asia Pacific (−108.2 million), Northeast Asia (−73.5 million), South Asia (−43 million), and Southeast Asia (−31.8 million). The total impact of COVID-19 on the travel and tourism labor market under various scenarios will depend, to a large extent, on the government's economic response to fight the virus.

Under the first and the second scenarios, governments’ economic response is essential but still not a last resort; under scenario 3, governments’ support to fight COVID-19 economic damages becomes crucial. Total jobless claims in the world could reach −514.8 million depending on two main factors: (1) if the entire tourist season is lost and (2) governments’ strength and speed of response.

Tourists’ spending in the travel and tourism industry worldwide will experience a drop ranging from −1.5 US$ bn to −771.7 US$ bn (see Fig. 7 ).

Fig. 7.

Estimated potential (COVID-19) impact on tourists’ spending in travel and tourism industry worldwide.

Source: Authors’ calculations using their system dynamic model

On the world level, tourist spending will drop from −604.8 US$ bn (scenario 1) to −1.9 US$ trillion (scenario 3). Regions most affected (in absolute numbers) will be the Americas (−138.3 to −419.1 US$ bn), Asia Pacific (−235.7 to −673.4 US$ bn), Europe (−192.9 to −771.7 US$ bn), Northeast Asia (−115.3 to −329.5 US$ bn), Central Asia (−1.5 to −4.2 US$ bn), North Africa (−9.0 to – 29.1 US$ bn), and sub-Saharan Africa (−11.8 to −38.2 US$ bn)—and these regions have the lowest absolute numbers in tourist spending. All other regions, as we can see from the tables, ranging from Central Asia (−1.5 US$ bn) to Europe (−771.7 US$ bn). Scenario 1 (as an actual scenario) shows the full extent of the pandemic crisis for the travel and tourism industry worldwide. Scenarios 2 and 3 show a profound impact on the crisis if pandemic outbreaks continue. On the world level, −1.3 US$ trillion in tourist spending will be lost under scenario 2 and −1.89 US$ trillion under scenario 3. The region most affected by the pandemic crisis will be the Asia Pacific due to its dependence on the travel and tourism industry. Policymakers and tourism practitioners can observe that scenario 1 is an affordable scenario for the travel and tourism industry. Under this scenario, the industry could recover from the crisis (regain lost income) in 15 months. However, scenarios 2 and 3 do not offer such optimistic conditions. Scenario 2 presents the resilience threshold, the point beyond which the travel and tourism industry will need massive government bailouts and incentives to recover from the crisis. Scenario 3 is the worst scenario that could take the travel and tourism industry back to the income levels of 2009 or even 1980. The third scenario would need a massive government support plan for the travel and tourism industry, similar to the one developed for the financial industry during the great recession of 2008.

Fig. 8 shows the impact of COVID-19 on investment flow in the tourism industry worldwide. Investment flow in the tourism industry will face large adverse shocks because of the negative expectations of the industry, and tourist practitioners do not have many choices to alter these conditions. Experiences from SARS (2002) and H1N1(2009) show that a recovery period in the number of tourist arrivals beyond the pandemic crisis and travel constraints lift-off last, on average, one year. Under such conditions, planning new investments in the industry is not expected. Another negative aspect is freezing investments and postponing future investments, since the recovery phase will involve price competition, not diversification of tourism supply (re-branding, innovation). Tourism practitioners should rewrite investment plans for the next two years, linking investments (if any) to price competition during the recovery phase and to product competition afterward.

Fig. 8.

Estimated potential (COVID-19) impact on capital investment in the travel and tourism industry worldwide.

Source: Authors’ calculations using their system dynamic model

As shown in Fig. 8, the world level travel and tourism industry will face a massive decrease in the total capital investments, and total capital investments in the travel and tourism industry on the world level will drop by −362.9 US$ bn under scenario 1. Regions with the most significant decline as the one with the highest inbound arrivals and share in the GDP. The highest drop of capital investments is for the Americas (−106.4 US$ bn) and Asia Pacific (−188.5), followed by Northeast Asia (−92.9), Northern America (−83.9), and Europe (−83.4). Such a massive decline, even under the best scenario 1, will send the tourism industry back to the total capital investment levels of 2004. Scenario 2 estimates that travel and tourism industry capital investments could drop to the 1989/1990 levels. Since capital investments are essential for tourism growth, a decrease in capital investment will also result in the decline of future inbound arrivals. According to our model, capital investments on the world level will decline by 781.5 US$ bn. The decline under scenario 3 is even more significant, reaching −1.1 US$ trillion of lost capital investments in the travel and tourism industry. The tourism industry will have to abandon innovation and product development, facilitate touring visitors, experience development, and mobilize efforts to deliver consumer experience through increasing traveler confidence and reducing perceived traveling risks.

In the short run, the total capital investment decline across regions will reshape the tourism industry worldwide. The level of strength in the restructuring process will depend on the pandemic dynamics and the scenarios in place. Capital investments in the travel and tourism industry show a high level of volatility. Risk condition realization, in the form of exogenous factors (pandemic outbreaks, terrorism, environmental disasters) or endogenous factors (financial and business cycles), will result, on average, in a −15 to −20 percentage points drop in the level of capital investments in the tourism industry. The recovery period for the investments, unlike the conditions in tourist arrivals or spending, is long, lasting on average two years to return to positive figures and 8–9 years to return to pre-crisis levels. The larger the plunge in capital investments, the longer the recovery period and convergence time to pre-crisis levels.

Model results (scenarios 1–3) show that aggregate indicators in the travel and tourism industry will register a significant fall in 2020. Tourism industry competitiveness and resilience will be tested as never before and will require meaningful public and private efforts to recover from the COVID-19 pandemic outbreak.

6. Discussion

Using PSVAR on data from 1995 to 2019 and system dynamic modeling (real-time data parameters connected to COVID-19), we estimated the potential impact of the current pandemic crisis on the tourism industry worldwide. Empirical studies on pandemic outbreaks and their impact on the tourism industry are not found in the literature. The COVID-19 pandemic outbreak will have an unmatched negative impact (vast decline in inbound tourism arrivals) on the travel tourism industry worldwide, and we empirically investigated those impacts in this study. Negative shocks will be significant, not just in the short run but also in the long run, and it will take several years for the industry to recover. The summary of our empirical results on a world level, demonstrated in the various scenarios, shows that the travel and tourism industry's contribution to the GDP will decline from −4.1 US$ trillion to −12.8 US$ trillion. In addition, the total tourism industry contribution to employment will fall from −164.506 million to −514.080 million jobs, and lost inbound tourist spending will plunge from −604.8 US$ bn to −1.9 US$ trillion with a fall in capital investments of −362.9 US$ bn to −1.1 US$ trillion.

Unlike past pandemic crises with idiosyncratic shocks having dominant effects, COVID-19, because of the travel restrictions and border closures, reveals large commons shocks (globalization effect) on the domestic tourism industry. During previous pandemic crises, like SARS (2002) and H1N1(2009), the domestic tourism industry suffered from idiosyncratic (domestic) shocks. Once pandemic cases were no longer registered, the tourist industry started to return (bouncing effect)—the extent of which depended on the risk perception, risk aversion, income levels, and hysteresis (Mao et al., 2010). Past studies note the high resilience of the tourism industry to shocks (WTTC, 2019), and the number of months needed to recover in the tourism sector decreased from 26 to 10 months on average from 2001 to 2018. However, this time it could be different.

Past pandemic crises operated mostly through idiosyncratic shocks' channels, exposing domestic tourism sectors to large negative shocks. Once domestic shocks disappeared (zero infection cases), inbound arrivals started to revive immediately. With the COVID-19 pandemic crisis, global effects in the form of common shocks multiply the intensity of the crisis. Thus, a country with no COVID-19 alert (idiosyncratic shocks) will not experience an immediate bounce-back effect of inbound arrivals if other countries do not withdraw the COVID-19 alert (common shock). For example, a COVID-19 alert can be lifted for the European region, but arrivals from China will not revive if the alert is not lifted for China as well. Pandemic crises such as COVID-19 show multiplier effects through both idiosyncratic and common shock channels, resulting in a much deeper crisis compared to past pandemic episodes. Past pandemic episodes were limited to idiosyncratic shocks and constrained common shocks. Now, the interaction of both idiosyncratic and common shocks is stretching the tourism industry to the limits. Another important interaction is with the financial cycles. Pandemic cycles appearing at the peak of financial cycles, as in this time, intensify the negative pandemic impacts, limiting the economic response of business and government. Capital investments and employment plunge, real wage and household income abruptly drop, resulting in a decline in aggregate consumption. In the condition of high private and public outstanding debt, both private initiative efforts in the tourism sector and the government's economic policy instruments to revive the industry are limited.

Krueger et al. (2020) show that the rational relocation of economic activities across sectors is a strong mitigation force, even though the government is not explicitly intervening. According to the findings of the Swedish Model Solution, agents need to adjust their sectoral actions independently to ensure that the economic and human costs of the COVID-19 crisis are substantially modulated without government interference.

While idiosyncratic shocks caused by pandemic episodes in the past disrupted the domestic tourism industry, the economy's aggregate consumption remained robust to a point (relying on other sectors). With COVID-19 causing global lockdowns, disrupting circular flows and economic transmission channels, both the tourism industry (idiosyncratic shock) and the rest of the economy (common shock) are under stress. It is the first case in modern times that a pandemic episode caused a global worldwide economic disruption on this scale. It starts with a crisis in the tourism sector and amplifies through financial cycles (current levels of private and public outstanding debt).

Policy must be resilient in the face of change and/or adaptive to various eventualities. In order to avoid this relatively short-term occurrence from having a long-term "scarring" impact, a major priority should be an economic policy response that supports the companies concerned. In addition, helping people who lose their income is important as well continuing to provide the needed public services. Eventually, policy makers will examine whether broader measures will be necessary to help employees who are losing their jobs or facing pay cuts (Emerson and Johnson, 2020; McKee and Stuckler, 2020). The environmental hazards impact on tourism (Halkos and Zisiadou, 2020) shows expected heavy economic losses for the most developed countries and significant deaths for the least developed countries.

7. Conclusion

Policy makers and practitioners in the tourism industry must develop a new crisis-readiness mechanism to fight the current pandemic crisis as well as future pandemic crises. To do so, they must gain empirical knowledge on the nature and actual extent of the COVID-19 crisis. For now, this has not happened, and scenarios developed by them significantly underestimate the potential effects of the COVID-19 pandemic crisis. Kirby (2020) recalls that central banks expect rapid tightening, representing the sharp fall in sovereign bond yields worldwide. Various countries are offering different economic assistance programs.

Policy makers and practitioners in the tourism industry need to gain knowledge of the impact of the pandemic crisis on the tourism industry and economy. In this study, we acquire the same knowledge by examining the historical effects of past pandemic outbreaks corrected the real-time parameters of COVD-19. A four-part economic strategy is required: (1) accept economic losses, (2) protect health, (3) support people experiencing a sudden loss of income by broadening existing security network programs, and (4) protect productive capacity and use economic production capacity to the fullest extent possible as soon as the virus has diminished (Marron, 2020).

Our study demonstrates that pandemic crises have long-lasting negative effects on the tourism industry and economy. Estimated negative effects are far beyond those observed during past pandemic crises. Future pandemic crises should be dealt with promptly, and to do so, policy makers and practitioners need effective contingency plans. Our study shows that the pandemic effects of COVID-19 on the tourism industry share the effect of a common shock. A revival of the tourism industry worldwide will need cooperation rather than competition to minimize the costs of COVID-19. Mandel and Veetil (2020) estimated that global production decreased by 7% when only China locked down, but it decreased by 23% at the peak of the crisis when other countries implemented lockdown. As the shock propagates across the world economy, such immediate consequences are compounded due to buyer–seller ties. In the optimistic and unlikely scenario of an end to all lockdowns, the world economy takes about one quarter period to achieve a new balance. If partial lockdowns persist, recovery time will likely be considerably longer.

Our sample was limited to 185 countries divided across regions according to WTTC methodology. We used annual data and the PSVAR auto-regression model and system dynamics. Monthly data availability enabled the use of other time series modeling techniques, which improved empirical knowledge overall. Further studies should move in that direction and involve larger and longer time series samples to improve model accuracy and robustness. Our study is a modest contribution to the field of pandemic economics and tourism, and we hope it will encourage further research on this critical issue.

CRediT authorship contribution statement

Marinko Škare: Data curation, Conceptualization, Methodology, Software. Domingo Riberio Soriano: Supervision, Validation, Writing - review & editing. Małgorzata Porada-Rochoń: Visualization, Writing - original draft.

Biographies

M. Škare is Vice Rector for Research, arts and Cooperation, at the University of Juraj Dobrila in Pula. He was awarded the Juraj Dobrila University Prize and Istrian country (2013–2014) for the development of social sciences. He is editor-in-chief of the journal Economic Research, Taylor & Francis (Routledge group). He received a high honor degree from the Croatian parliament “National Science Award in 2014″ as recognition for exceptional research results. He serves on the editorial board of several internationally recognized journals and reviewers for WoS-ranked journals and as member of the Board of Governors of ACIEK. He is author of 140 scientific articles and majority of them in the ISI ranked journals.

Domingo Enrique Ribeiro-Soriano is Professor of Business Administration at the University of Valencia, Spain, Associate Editor of the Journal of Business Research, Elsevier, Senior editor of the European Journal of International Management, Inderscience, and Chair of the Cathedra Entrepreneurship: Being student to entrepreneur’ – Grupo Maicerías Españolas Arroz DACSA. He has published more than 100 papers in ISI ranked journals and has had a guest editing role in more than 25 issues of journals cited in the SSCI of Thomson Reuters-Clarivate. He has worked in Ernst & Young Consulting and was Director of European Community Programs.

Małgorzata Porada-Rochoń is associate Professor of Finanse at the University of Szczecin, Poland. Investator of several national and international research projects as well as author/coauthor of more than 50 publications. Expert in economic and financial evaluation of projects co-financed from EU funds.