1 Introduction

The coronavirus pandemic (COVID-19) has not only had a significant impact on public health, but it has also severely affected one of the linchpins of the European Union (EU) economy—the tourism sector. According to the latest data from Eurostat, 10% of the EU non-financial business economy belonged to the tourism industries, employing 12.3 million people; 75% of those industries operated in accommodation (14%) or food and beverage serving (61%) activities, while 55% of them were located in Italy, France, Spain and Germany.Footnote 1 As many countries have introduced curfews and travel restrictions to contain the spread of the coronavirus, international travels have come to an almost complete standstill. Moreover, the long-lasting lockdowns imposed in many countries also disturbed the supply chain of the tourism industry, since many tourism related activities were put on hold. The latest data from the World Tourism Organization (UNWTO) show that the international arrivals in Europe dropping by 70%: in Western Europe by 65%; in Southern Europe by 71%; in Northern Europe by 75%; and in Central Europe by 70%. More than a half of the loss of worldwide international arrivals (−1.07 billion persons) belongs to Europe (− 0.52 billion persons).Footnote 2 Thus, while the countries were seeing a novice drop in tourist arrivals, all the sectors that depend on tourism activities are facing similar setbacks, ranging from closures in the food services to decreasing guest numbers in the accommodation (see e.g., Škare 2020).Footnote 3

During the pandemic, a significant number of studies explore the decline of tourism activities and the corresponding impact on the economic system (see, e.g., Farzanegan et al. 2020; Lee and Chen 2020; Mariolis et al. 2020; Qiu et al. 2020; Tsionas 2020; Yang et al. 2020). The purpose of this paper is to provide estimations for the COVID-19 tourism multiplier effects on output, employment and trade balance of the German and Spanish economies. Since tourism has become an important part of economic activities during the last decades, it is reasonable to expect that the COVID-19 shock on the tourism sector will significantly affect the economic system due to (a) the loss of jobs in tourism, which will reduce incomes of the people involved in the tourism industry and, therefore, will diminish aggregate consumption; and (b) the fall in demand for intermediate inputs by the sectors that supply inputs to the tourism industry. Thus, it is important to study the intersectoral effects of the decline in tourism activities. For this purpose, we follow the approach of Mariolis et al. (2020), i.e., we use an analytic framework inspired by the concept of the Sraffian multiplier (Kurz 1985; Metcalfe and Steedman 1981; Mariolis 2008a) and data from the Symmetric Input Output Tables (SIOTs) for the year 2015 (latest data at the time of this research) provided via the Organisation for Economic Co-operation and Development (OECD) website, https://stats.oecd.org. As it has been demonstrated, the multiplier for real-world economies does not constitute a scalar, but a matrix quantity reflecting, in a complex way, the underlying inter-industry socio-technical linkages. It could, furthermore, be shown that the Sraffian multiplier includes as special versions or limit cases the usual Keynesian multiplier, the multipliers of the traditional input–output analysis, and their Marxian versions.Footnote 4 Since Germany (Spain) is characterized by a relatively low (high) dependency on the tourism sector, this empirical analysis will reveal whether the differences in the downturn within the EU economy can be attributed to the degree of the dependency of the economies on tourism activities.Footnote 5

The remainder of the article is structured as follows. The second section presents the method. The third section presents the empirical estimations. The fourth section concludes.

2 The analytical framework

We consider an open, linear economy involving only single products, “basic” commodities (in the sense of Sraffa 1960, pp. 7–8). In addition, we assume that (i) all capital is circulating; (ii) the input–output coefficients are fixed; (iii) there are non-competitive imports; (iv) the net product is distributed to profits and wages that are paid at the end of the common production period; (v) the price of a commodity obtained as an output at the end of the production period is the same as the price of that commodity used as an input at the beginning of that period (“stationary prices”); and (vi) labour is homogeneous within each industry but heterogeneous across industries.Footnote 6

On the basis of these assumptions, the price side of the system is described byFootnote 7

$$\textbf{p} = \textbf{pA}[\textbf{I} + \hat{\textbf{r}}] + \textbf{w}\hat{\textbf{l}}$$
(1)

where \(\bf p\) (\(> \textbf{0}\)) the \(1 \times n\) vector of commodity prices, \(\bf A\) (\(\ge \textbf{0}\)) the \(n \times n\) matrix of total input–output coefficients, \(\bf I\) the \(n \times n\) identity matrix, \(\hat{\textbf{r}}\) (\(r_{j} > - 1\) and \(\hat{\textbf{r}} \ne \textbf{0}\)) the \(n \times n\) diagonal matrix of the sectoral profit rates, \(\bf w\)(\(w_{j} > 0\)) the \(1 \times n\) vector of money wage rates, \(\hat{\textbf{l}}\) (\(l_{j} > 0\)) the \(n \times n\) diagonal matrix of direct labour coefficients.

The quantity side of the system is described by

$$\textbf{x}^{T} = \textbf{Ax}^{T} + \textbf{y}^{T}$$

and

$$\textbf{y}^{T} = \textbf{c}_{w}^{T} + \textbf{c}_{p}^{T} - \textbf{Im}^{T} + \textbf{d}^{T}$$
(2)

where \(\textbf{x}^{T}\) denotes the gross output vector, \(\textbf{y}^{T}\) the vector of net output, \(\textbf{c}_{w}^{T}\) (\(\textbf{c}_{p}^{T}\)) the vector of consumption demand out of wages (profits), \(\textbf{Im}^{T}\) the import demand vector, and \(\textbf{d}^{T}\) the autonomous demand vector (government expenditures, investments and exports).

By taking into account Eqs. (1) and (2), we derive the following equationFootnote 8:

$$\textbf{y}^{T} = \Pi \textbf{d}^{T}$$
(3)

where \(\Pi \equiv [\textbf{I} - \textbf{C} + \textbf{M}]^{ - 1}\) denotes the \(n \times n\) matrix of multipliers linking autonomous demand to net output. Furthermore, \(\textbf{C} \equiv (\textbf{pc}^{T} )^{ - 1} \textbf{c}^{T} [\textbf{p} - (s_{w} \textbf{w}\Lambda + s_{p} \textbf{p}\tilde{\textbf{H}})]\) denotes the \(n \times n\) matrix of total consumption demand, cT the uniform consumption pattern (associated with the two types of income), \(s_{w}\) (\(s_{p}\)) the saving ratio out of wages (profits), \(\Lambda \equiv \hat{\textbf{l}}[\textbf{I} - \textbf{A}]^{ - 1}\) the \(n \times n\) matrix of “vertically integrated labour coefficients”, and \(\tilde{\textbf{H}} \equiv \textbf{A}\hat{\textbf{r}}[\textbf{I} - \textbf{A}]^{ - 1}\) the \(n \times n\)\(\hat{\textbf{r}} -\) vertically integrated technical coefficients matrix”. Finally, \(\textbf{M} \equiv \hat{\textbf{m}}[\textbf{I} - \textbf{A}]^{ - 1}\) denotes the \(n \times n\) matrix of total import demand, and \(\hat{\textbf{m}}\) the \(n \times n\) diagonal matrix of imports per unit of gross output of each commodity. Therefore, the multiplier effects depend, in a rather complicated way, on the: (i) technical conditions of production; (ii) imports per unit of gross output; (iii) income distribution; (iv) savings ratios out of wages and profits; (v) consumption pattern; and (vi) physical composition of autonomous demand.

From Eq. (3) and given that \(\textbf{L}^{T} \equiv \hat{\textbf{l}}\textbf{x}^{T}\) denotes the vector of sectoral employment, we derive the equation:

$$\textbf{L}^{T} = \Lambda \Pi \textbf{d}^{T}$$
(4)

where \(\Lambda \Pi\) denotes the \(n \times n\) matrix of employment multipliers linking autonomous demand to total employment.

Furthermore, we derive the matrix multiplier linking autonomous demand to imports as

$$\textbf{Im}^{T} = \hat{\textbf{m}}[\textbf{I} - \textbf{A}]^{ - 1} \Pi \textbf{d}^{T}$$
(5)

where \(\hat{\textbf{m}}[\textbf{I} - \textbf{A}]^{ - 1} \Pi\) denotes the \(n \times n\) matrix of import multipliers linking autonomous demand to imports.

Finally, in the closed economy case, i.e., for \(\hat{\textbf{m}}\)\(= \textbf{0}\) matrix of multipliers reduces to \(\Pi_{c} \equiv [\textbf{I} - \textbf{C}]^{ - 1}\) and thus Eqs. (3) and (4) become

$$\textbf{y}^{T} = \Pi_{c} \textbf{d}^{T}$$
(6)
$$\textbf{L}^{T} = \Lambda \Pi_{c} \textbf{d}^{T}$$
(7)

Now, to estimate the multiplier effects of international travel receipts on output, total employment and imports, we set \(\textbf{d}^{T}\) equal to the vector that gives the distribution of international travel receipts to the 36 sectors that are described in the SIOTs of the economy of Germany and Spain.Footnote 9 The elements of dT are weighed by the total international travel receipts of the two economies and, therefore, the sum of the elements of dT equal to 1. The allocation of international travel receipts for the German economy to the commodities is depicted in Fig. 1: about 69.5% of the international travel receipts of the German economy corresponds to the commodities “Accommodation and food services” (46.3%), “Education” (17.7%), and “Transportation and storage” (5.6%), while about 30.5% of the travel receipts is distributed to the rest of the 32 commodities. The allocation of international travel receipts for the Spanish economy to the commodities is depicted in Fig. 2: about 70.0% of the international travel receipts of the Spanish economy corresponds to the commodities “Accommodation and food services” (44.4%), “Transportation and storage” (14.6%), “Arts” (5.7%), and “Real estate services” (5.3%), while about 30.5% of the travel receipts is distributed to the rest of the 31 commodities.

Fig. 1
figure 1

Distribution (%) of international travel receipts per commodity, Germany 2015

Fig. 2
figure 2

Distribution (%) of international travel receipts per commodity, Spain 2015

3 Empirical analysis

We apply the previous analysis to the SIOT’s of the German and Spanish economies for the case, where \(s_{w} = 0\) and \(s_{p} = 1\).Footnote 10 The empirical results indicate that a decrease in international travel receipts by 1 million euros would lead to a total (direct and indirect)Footnote 11:

  1. 1.

    Decrease in net output of about 1.25 (1.33) million euros in the German (Spanish) economy

  2. 2.

    Decrease in employment of about 21.57 (21.93) people in the German (Spanish) economy.

  3. 3.

    Decrease in imports of about 0.42 (0.30) million euros in the German (Spanish) economy.

These findings indicate that the tourism multiplier effects are more significant in the Spanish than in the German economy, in the sense that Spain is characterized by a higher output and employment tourism multiplier and a lower import tourism multiplier. Nevertheless, in both economies, tourism can be characterized as a key-sector, in the sense that the output and employment tourism multipliers are above the economies’ average multipliers and, at the same time, the import tourism multipliers are below the average import multipliers. Table 1 presents the average multipliers for the German and Spanish economies in comparison with the respective tourism multipliers.

Table 1 Average multipliers for the German and Spanish economies

Now, according to OECD, in the year 2019, Germany’s GDP accounted for 4,153,928.4 million US dollars and total employment reached 42,400,100 persons.Footnote 12 Moreover, according to World Tourism Organization (UNWTO), in the year 2019, Germany’s international travel receipts reached 41.8 billion US dollars, while according to the latest available data travel receipts declined by 46% in the year 2020.Footnote 13 From our analysis of the multiplier effects of international travel receipts on the German economy, it then follows that this loss of travel receipts corresponds to a decrease in GDP of about 0.58%, a decrease in the levels of employment of about 0.98%, and a decrease in the surplus of the balance of goods and services of about 11.2 billion US dollars (19.228 billion US dollars decrease in exports minus 8.076 billion US dollars decrease in imports), respectively. In the polar (and hypothetical) case, where all the travel receipts (41.8 billion US dollars) are lost, our analysis indicates that GDP and employment in the German economy would decrease by 1.26% and 2.13%, respectively, while the surplus of the balance of goods and services would decrease by 24.2 billion US dollars. According to same data sources, in the year 2019, Spain’s GDP accounted for 1,796,023.9 million US dollars, total employment reached 19,779,300 persons, international travel receipts reached 79.7 billion US dollars, while according to the latest available data travel receipts declined by 77% in the year 2020. From our analysis of the multiplier effects of international travel receipts on the Spanish economy, it then follows that this loss of travel receipts corresponds to a decrease in GDP of about 4.54%, a decrease in the levels of employment of about 6.80%, and a decrease in the surplus of the balance of goods and services of about 43.0 billion US dollars (61.369 billion US dollars decrease in exports minus 18.411 billion US dollars decrease in imports), respectively. Finally, in the polar case, where all the travel receipts (79.7 billion US dollars) are lost, our analysis indicates that GDP and employment in the Spanish economy would decrease by 5.90% and 8.84%, respectively, while the surplus of the balance of goods and services would decrease by 55.8 billion US dollars. Thus, given that Germany’s (Spain’s) GDP decreased by 4.9% (11%) in the year 2020, our estimations imply that the decline in inbound tourism accounts for about 11.8% (41.3%) of the observed recession in the German (Spanish) economy.

To further delve into the differences between the tourism multiplier effects in Germany and Spain, we also estimate the economies’ average and tourism multipliers for the closed economy case (Eqs. 6 and 7). Table 2 presents the closed-economy average multipliers for the German and Spanish economies in comparison with the respective tourism multipliers.

Table 2 Closed-economy average multipliers for the German and Spanish economies

We observe that, in the closed-economy case, Germany is characterized by a higher output and employment multiplier of the tourism sector and by higher average output multiplier than Spain, while Spain is characterized by a higher average employment multiplier. Moreover, it is interesting to note that, in the closed-economy case, the tourism sector can be characterized as key-sector (anti-key-sector) for the German (Spanish) economy, since the output and employment multipliers of tourism are higher (lower) than the economy’s average.

Moreover, given that our estimates are based on the assumption that all wages are consumed and all profits are saved, it follows that the higher the wages paid per unit of output, \(\textbf{w}\Lambda\), the higher the multiplier effect. Table 3 reports the average wages paid per unit of output for each economy as well as the wages paid per unit of tourism output, i.e., \(\textbf{w}\Lambda \textbf{d}^{T} ( = 1 - \textbf{p}\tilde{\textbf{H}}d^{T} )\) for the case, where \(\textbf{d}^{T}\) is set equal to the vector that gives the distribution of international travel receipts.

Table 3 Wages per unit of output in the German and Spanish economies

We observe that Germany is characterized by relatively higher wages per unit of output, and that the tourism sector of both countries is characterized by higher wages per unit of output compared to the respective economy’s average.

From the above, it follows that the relatively higher impact of the decline of international travel receipts in the Spanish economy than in the German economy can be attributed to (i) the relatively lower import multiplier of tourism in the Spanish economy, which leads to a higher open-economy output multiplier of tourism in the Spanish economy (Table 1); (ii) the relatively higher share of international travel receipts in total demand for the Spanish economy, which leads to higher total effects in the Spanish economy than in the German economy (1.26% decline of GDP in the German economy in the polar scenario of a 100% loss in travel receipts, and 5.90% decline of GDP in the Spanish economy, respectively); and (iii) the relatively higher decline of international travel receipts in the Spanish economy than in the German economy in the year 2020 (77% decline of international travel receipts in the Spanish economy, and 46% decline in the German economy).

Since it is not only important to focus on the aggregate effects of the decline in tourism activities but also on the intersectoral dimension of these effects, in Figs. 3 and 4, we present the distribution of the decrease in net output per commodity in the German (Spanish) economy; in Figs. 5 and 6, we present the distribution of the decrease in employment per sector in the German (Spanish) economy, and in Figs. 7 and 8, we present the distribution of the decrease in imports per commodity in the German (Spanish) economy.

Fig. 3
figure 3

Distribution (%) of the output multiplier, Germany 2015

Fig. 4
figure 4

Distribution (%) of the output multiplier, Spain 2015

Fig. 5
figure 5

Distribution (%) of the employment multiplier, Germany 2015

Fig. 6
figure 6

 Distribution (%) of the employment multiplier, Spain 2015

From Figs. 3, 5, and 7, it is deduced that about 55.9% of net output losses for the German economy correspond to “Accommodation and food services” (29.4%), “Education” (14.2%) and “Real estate services” (12.3%), about 55.5% of the decreases in the levels of employment correspond to the sectors “Accommodation and food services” (33.4%), “Education” (11.5%) and “Wholesale and retail trade” (10.6%), while the most significant decreases in imports correspond to “Accommodation and food services” (27.9%), “Wholesale and retail trade” (9.7%) and “Food products” (8.1%). From Figs. 4, 6, and 8, it is deduced that about 60.3% of net output losses for the Spanish economy correspond to “Accommodation and food services” (37.2%), “Real estate services” (12.2%) and “Transportation and storage” (10.9%), about 52.5% of the decreases in the levels of employment correspond to the sectors “Accommodation and food services” (27.7%), “Wholesale and retail trade” (15.7%) and “Other business sector services” (9.1%), while the most significant decreases in imports correspond to “Wholesale and retail trade” (12.2%), “Mining products” (10.7%) and “Food products” (9.7%).

Fig. 7
figure 7

 Distribution (%) of the import multiplier, Germany 2015

Fig. 8
figure 8

Distribution (%) of the import multiplier, Spain 2015

From the above, it follows that “Education” is a significant part of international travel receipts for Germany, both in terms of output and employment, while this is not true for the Spanish economy. This fact indicates that educational tourism is an important part of the German tourism industry. In addition, the relative importance of “Transportation and storage” is more significant, in terms of output and employment for the German than the Spanish economy. Finally, it is worth-mentioning that the “Accommodation and food services” is a significant part of the tourism import multiplier of the German economy but not for the Spanish economy.

It goes without saying that, although in this paper we focus on the ceteris paribus effects from a change in international travel receipts, this analytical framework can incorporate any shocks in the elements of autonomous demand and provide estimations for a vast range of alternative scenarios. For instance, the increase in government spending had a positive counter-impact on the economies (regarding both output and employment). In fact, our findings suggest that an increase in government spending by 1 million euros leads to a total (direct and indirect) increase in net output of about 1.53 (1.59) million euros in the German (Spanish) economy; an increase in employment of about 21.28 (27.47) persons in the German (Spanish) economy; and an increase in imports of about 21.28 (27.47) persons in the German (Spanish) economy. Thus, it follows that the multiplier effects of government spending on both economies are stronger than those of the tourism sector.Footnote 14 Therefore, well-targeted fiscal policies and the implementation of appropriate sectoral redistributive policies could significantly offset some of the negative COVID-19 shocks.

4 Concluding remarks

This paper provided estimations of the effects of a decrease in international travel receipts on output, total employment and trade balance of the German and Spanish economies, using a multisectoral model of single production and data from the SIOT’s for the year 2015. Based on the available facts and figures, it has been estimated that the decrease of international travel receipts recorded in the year 2020 in the German (Spanish) economy corresponds to a decrease in GDP of about 0.58% (4.54%), a decrease in the levels of employment of about 0.98% (6.80%), and a decrease in the surplus of the balance of goods and services of about 11.2 (43.0) billion US dollars, respectively. These decreases mainly affect the sectors “Accommodation and food services”, “Education”, “Real estate services”, and “Wholesale and retail trade” of the German economy, and the sectors “Accommodation and food services”, “Real estate services”, “Transportation and storage”, and “Wholesale and retail trade” of the Spanish economy.

Although in both countries tourism can be characterized as a key-sector for effective demand management policy, the tourism multiplier effects are more significant in the Spanish than in the German economy, which is also reflected in our estimations for the contribution of the decline in international travel receipts to the observed recession in the two countries. Since Germany (Spain) can be considered as a sufficiently representative economy of a relatively low (high) dependency on the tourism sector, our results seem to be in accordance with the relative higher observed recession in the Southern European Economies, which are more dependent on the highly vulnerable in the pandemic tourism industry.

This modelling can easily incorporate a wide range of estimations for different countries, involving alternative scenarios for all the components of autonomous demand that hopefully will contribute to both the better understanding of the intersectoral dimensions of the decline in economic activities by the spread of COVID-19 and the formulation of well-targeted recovery programs. Furthermore, future research efforts could extend the current framework to explore the COVID-19 international spillover effects by means of multi-country models and relevant inter-country input–output data.